You are here

Lo, Joseph Yuan-Chieh

Overview:

My lab focuses on the diagnosis and treatment of breast cancer using advanced imaging techniques. There are 3 main projects: tomosynthesis imaging, radiomics, and breast modeling.

First, I lead a team from the Ravin Advanced Imaging Laboratories (RAI Labs) that collaborated closely with Siemens Healthcare to develop digital breast tomosynthesis (DBT) imaging, a form of limited-angle tomography also known as "3D mammography." DBT can acquire a 3D image quickly, easily, and at comparable dose to conventional mammography. By improving both sensitivity and specificity of breast cancer diagnosis, DBT has become the most exciting recent development in breast cancer screening, and the only technology with the potential to replace mammography in the near future. This work led to the FDA approval of the Siemens DBT system. We continue to investigate DBT in terms of clinical protocols and physics optimization. 

Second, radiomics is an interdisciplinary field combining computer vision, machine learning, and informatics. We developed computer vision algorithms to detect suspicious mammographic lesions. We also created predictive models that use machine learning and statistical analysis in order to classify mammograms as benign versus malignant. In ongoing studies funded by NIH and DOD, we are addressing the clinically significant challenge of over-diagnosis of DCIS. By exploring the relationship between imaging findings and genomic markers, we hope to predict which cases of DCIS are likely to be indolent vs. aggressive, thus providing women with more personalized risk assessment to inform their treatment decisions.  

Finally, we are designing new virtual breast models that are based on actual patient data. These models go far beyond conventional phantoms in portraying realistic breast anatomy. Furthermore, we can transform these virtual models into physical form using the latest 3D printing technology. Such physical phantoms can be scanned on actual mammography and DBT systems, allowing us to measure image quality in new ways that are not only quantitative but also clinically relevant. We continue to refine the realism of these physical phantoms, and seek to develop new procedures for quality control, system evaluation, and the long term goal of virtual clinical trials.

Positions:

Professor of Radiology

Radiology
School of Medicine

Professor of Biomedical Engineering

Biomedical Engineering
Pratt School of Engineering

Member of the Duke Cancer Institute

Duke Cancer Institute
School of Medicine

Education:

B.S.E.E. 1988

B.S.E.E. — Duke University

Ph.D. 1993

Ph.D. — Duke University

Grants:

Breast Cancer Detection Consortium

Administered By
Surgery, Surgical Sciences
AwardedBy
National Institutes of Health
Role
Co Investigator
Start Date
September 19, 2016
End Date
August 31, 2021

Genomic Diversity and the Microenvironment as Drivers of Progression in DCIS

Administered By
Surgery, Advanced Oncologic and Gastrointestinal Surgery
AwardedBy
Department of Defense
Role
Co Investigator
Start Date
September 30, 2014
End Date
September 29, 2020

Machine learning and collaborative filtering tools for personalized education in digital breast tomosynthesis

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Investigator
Start Date
September 01, 2016
End Date
May 31, 2020

Training in Medical Imaging

Administered By
Biomedical Engineering
AwardedBy
National Institutes of Health
Role
Mentor
Start Date
July 15, 2003
End Date
August 31, 2019

(PQC3) Genomic Diversity and Microenvironment as Drivers of Metastasis in DCIS

Administered By
Surgery, Advanced Oncologic and Gastrointestinal Surgery
AwardedBy
National Institutes of Health
Role
Co Investigator
Start Date
August 01, 2014
End Date
July 31, 2018

Molecular and Radiologic Predictors of Invasion in a DCIS Active Surveillance Cohort

Administered By
Surgery, Advanced Oncologic and Gastrointestinal Surgery
AwardedBy
Breast Cancer Research Foundation
Role
Co Investigator
Start Date
October 01, 2016
End Date
September 30, 2017

Improved education in digital breast tomosynthesis using machine learning and computer vision tools

Administered By
Radiology
AwardedBy
Radiological Society of North America
Role
Advisor
Start Date
July 01, 2014
End Date
March 31, 2016

(PQA5) 'Dose and Mechanisms of Exercise in Breast Cancer Prevention'

Administered By
Radiation Oncology
AwardedBy
National Institutes of Health
Role
Co Investigator
Start Date
September 23, 2013
End Date
February 14, 2014

3D Digital Breast Phantoms For Multimodality Research

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Investigator
Start Date
January 01, 2010
End Date
January 31, 2014

Cross-disciplinary Training in Medical Physics

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Mentor
Start Date
July 01, 2007
End Date
June 30, 2013

Information-Theoretic Based CAD in Mammography

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Scientist
Start Date
July 01, 2003
End Date
June 30, 2011

Tomosynthesis for Improved Breast Cancer Detection

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Principal Investigator
Start Date
June 20, 2006
End Date
April 30, 2011

Accurate Models for Predicting Radiation-Induced Injury

Administered By
Radiation Oncology
AwardedBy
National Institutes of Health
Role
Investigator
Start Date
May 10, 2006
End Date
April 30, 2011

Reducing Benign Breast Biopsies with Computer Modeling

Administered By
Radiology, Breast Imaging
AwardedBy
National Institutes of Health
Role
Co Investigator
Start Date
June 01, 2003
End Date
May 31, 2008

Predicting Breast Cancer With Ultrasound and Mammography

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Principal Investigator
Start Date
March 01, 2002
End Date
February 28, 2005

Improved Diagnosis of Breast Microcalcification Clusters

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Principal Investigator
Start Date
July 01, 2001
End Date
June 30, 2004

Computer-Aided Diagnosis Of Breast Cancer Invasion

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Principal Investigator
Start Date
July 01, 1998
End Date
June 30, 2003

Computer Aid for the Decision to Biopsy Breast Lesions

Administered By
Radiology
AwardedBy
US Army Medical Research
Role
Co Investigator
Start Date
July 01, 1999
End Date
June 30, 2002

Computer Aid for the Decision to Biopsy Breast Lesions

Administered By
Radiology
AwardedBy
National Institutes of Health
Role
Investigator
Start Date
September 01, 1999
End Date
August 31, 2001
Show More

Publications:

A novel physical anthropomorphic breast phantom for 2D and 3D x-ray imaging.

Physical phantoms are central to the evaluation of 2D and 3D breast-imaging systems. Currently, available physical phantoms have limitations including unrealistic uniform background structure, large expense, or excessive fabrication time. The purpose of this work is to outline a method for rapidly creating realistic, inexpensive physical anthropomorphic phantoms for use in full-field digital mammography (FFDM) and digital breast tomosynthesis (DBT).The phantom was first modeled using analytical expressions and then discretized into voxels of a specified size. The interior of the breast was divided into glandular and adipose tissue classes using Voronoi segmentation, and additional structures like blood vessels, chest muscle, and ligaments were added. The physical phantom was then fabricated from the virtual model in a slice by slice fashion through inkjet printing, using parchment paper and a radiopaque ink containing 33% (I33% ) or 25% (I25% ) iohexol by volume. Three types of parchment paper (P1, P2, and P3) were examined. The phantom materials were characterized in terms of their effective linear attenuation coefficients (μeff ) using full-field digital mammography (FFDM) and their energy-dependent linear attenuation coefficients (μ(E)) using a spectroscopic energy discriminating detector system. The printing method was further validated on the basis of accuracy, print consistency, and the reproducibility of ink batches.The μeff of two types of parchment paper were close to that of adipose tissue, with μeff = 0.61 ± 0.05 cm-1 for P1, 0.61 ± 0.04 cm-1 for P2, and 0.57 ± 0.03 cm-1 for adipose tissue. The addition of the iodinated ink increased the effective attenuation to that of glandular tissue, with μeff = 0.89 ± 0.06 cm-1 for P1 + I25% and 0.94 ± 0.06 cm-1 for P1 + I33% compared to 0.90 ± 0.03 cm-1 for glandular tissue. Spectroscopic measurements showed a good match between the parchment paper and reference values for adipose and glandular tissues across photon energies. Good accuracy was found between the model and the printed phantom by comparing a FFDM of the virtual model simulated through Monte Carlo with a real FFDM of the fully printed phantom. High consistency was found over multiple prints, with 3% variability in mean ink signal across various samples. Reproducibility of ink consistency was very high with <1% variation signal from multiple batches of ink. Imaging of the phantom using FFDM and DBT systems showed promising utility for 2D and 3D imaging.A novel, realistic breast phantom can be created using an analytically defined breast model and readily available materials. The work provides a method to fabricate any virtual phantom in a manner that is accurate, inexpensive, easily accessible, and can be made with different materials or breast models.

Authors
Ikejimba, LC; Graff, CG; Rosenthal, S; Badal, A; Ghammraoui, B; Lo, JY; Glick, SJ
MLA Citation
Ikejimba, LC, Graff, CG, Rosenthal, S, Badal, A, Ghammraoui, B, Lo, JY, and Glick, SJ. "A novel physical anthropomorphic breast phantom for 2D and 3D x-ray imaging." Medical physics 44.2 (February 2017): 407-416.
PMID
27992059
Source
epmc
Published In
Medical physics
Volume
44
Issue
2
Publish Date
2017
Start Page
407
End Page
416
DOI
10.1002/mp.12062

Assessing task performance in FFDM, DBT, and synthetic mammography using uniform and anthropomorphic physical phantoms.

The purpose of this study is to quantify the differences in detectability between full field digital mammography (FFDM), digital breast tomosynthesis (DBT), and synthetic mammography (SM) for challenging, low contrast signals, in the context of both a uniform and an anthropomorphic, textured phantom.Images of the phantoms were acquired using a Hologic Selenia Dimensions system. Images were taken at 50%, 100%, and 200% of the dose delivered under automatic exposure control (AEC). Low-contrast disks, created using an inkjet printer with iodine-doped ink, were inserted into the phantom. The disks varied in diameter from 210 to 630 μm, and in local contrast from 1.1% to 2.8% in regular increments. Human observers located the disks in a 4 alternative forced choice experiment. Proportion correct (PC) was computed as the number of correct localizations out of the total number of tries.Overall, scores from FFDM and DBT were consistently greater than scores from SM. At an exposure corresponding to the AEC setting, mean PC scores for the largest disks with the uniform phantom were 0.80 for FFDM, 0.83 for DBT, and 0.66 for SM, with the same rank ordering at other doses. Scores were similar but lower for the nonuniform background. At an exposure twice the AEC setting, however, the difference between uniform and nonuniform scores was most pronounced for DBT alone. Differences between scores for FFDM and SM were statistically significant, while those between FFDM and DBT were not. Scores were used to compute the minimum contrast level needed to reach 62.5% detection rate. The minimum contrast for SM was 36%-81% higher compared to FFDM or DBT, in either background.This study shows that an anthropomorphic phantom and lesions inserts may be used to conduct a reader study. Detectability was significantly lower for synthetic mammography than for FFDM or DBT, for all conditions. Additionally, observer performance was consistently lower for the anthropomorphic phantom, indicating the greater challenge due to anatomical background. Because of this, it may be important to use realistic phantoms in observer studies in order to draw conclusions that are more clinically relevant.

Authors
Ikejimba, LC; Glick, SJ; Choudhury, KR; Samei, E; Lo, JY
MLA Citation
Ikejimba, LC, Glick, SJ, Choudhury, KR, Samei, E, and Lo, JY. "Assessing task performance in FFDM, DBT, and synthetic mammography using uniform and anthropomorphic physical phantoms." Medical physics 43.10 (October 2016): 5593-.
PMID
27782687
Source
epmc
Published In
Medical physics
Volume
43
Issue
10
Publish Date
2016
Start Page
5593
DOI
10.1118/1.4962475

Predicting false negative errors in digital breast tomosynthesis among radiology trainees using a computer vision-based approach

Authors
Wang, M; Zhang, J; Grimm, LJ; Ghate, SV; Walsh, R; Johnson, KS; Lo, JY; Mazurowski, MA
MLA Citation
Wang, M, Zhang, J, Grimm, LJ, Ghate, SV, Walsh, R, Johnson, KS, Lo, JY, and Mazurowski, MA. "Predicting false negative errors in digital breast tomosynthesis among radiology trainees using a computer vision-based approach." Expert Systems with Applications 56 (September 2016): 1-8.
Source
crossref
Published In
Expert Systems with Applications
Volume
56
Publish Date
2016
Start Page
1
End Page
8
DOI
10.1016/j.eswa.2016.01.053

Impact of breast structure on lesion detection in breast tomosynthesis, a simulation study.

This study aims to characterize the effect of background tissue density and heterogeneity on the detection of irregular masses in breast tomosynthesis, while demonstrating the capability of the sophisticated tools that can be used in the design, implementation, and performance analysis of virtual clinical trials (VCTs). Twenty breast phantoms from the extended cardiac-torso (XCAT) family, generated based on dedicated breast computed tomography of human subjects, were used to extract a total of 2173 volumes of interest (VOIs) from simulated tomosynthesis images. Five different lesions, modeled after human subject tomosynthesis images, were embedded in the breasts and combined with the lesion absent condition yielded a total of [Formula: see text] VOIs. Effects of background tissue density and heterogeneity on the detection of the lesions were studied by implementing a composite hypothesis signal detection paradigm with location known exactly, lesion known exactly or statistically, and background known statistically. Using the area under the receiver operating characteristic curve, detection performance deteriorated as density was increased, yielding findings consistent with clinical studies. A human observer study was performed on a subset of the simulated tomosynthesis images, confirming the detection performance trends with respect to density and serving as a validation of the implemented detector. Performance of the implemented detector varied substantially across the 20 breasts. Furthermore, background tissue density and heterogeneity affected the log-likelihood ratio test statistic differently under lesion absent and lesion present conditions. Therefore, considering background tissue variability in tissue models can change the outcomes of a VCT and is hence of crucial importance. The XCAT breast phantoms have the potential to address this concern by offering realistic modeling of background tissue variability based on a wide range of human subjects, comprising various breast shapes, sizes, and densities.

Authors
Kiarashi, N; Nolte, LW; Lo, JY; Segars, WP; Ghate, SV; Solomon, JB; Samei, E
MLA Citation
Kiarashi, N, Nolte, LW, Lo, JY, Segars, WP, Ghate, SV, Solomon, JB, and Samei, E. "Impact of breast structure on lesion detection in breast tomosynthesis, a simulation study." Journal of medical imaging (Bellingham, Wash.) 3.3 (July 2016): 035504-.
PMID
27660807
Source
epmc
Published In
Journal of medical imaging (Bellingham, Wash.)
Volume
3
Issue
3
Publish Date
2016
Start Page
035504
DOI
10.1117/1.jmi.3.3.035504

Finite-element modeling of compression and gravity on a population of breast phantoms for multimodality imaging simulation.

The authors are developing a series of computational breast phantoms based on breast CT data for imaging research. In this work, the authors develop a program that will allow a user to alter the phantoms to simulate the effect of gravity and compression of the breast (craniocaudal or mediolateral oblique) making the phantoms applicable to multimodality imaging.This application utilizes a template finite-element (FE) breast model that can be applied to their presegmented voxelized breast phantoms. The FE model is automatically fit to the geometry of a given breast phantom, and the material properties of each element are set based on the segmented voxels contained within the element. The loading and boundary conditions, which include gravity, are then assigned based on a user-defined position and compression. The effect of applying these loads to the breast is computed using a multistage contact analysis in FEBio, a freely available and well-validated FE software package specifically designed for biomedical applications. The resulting deformation of the breast is then applied to a boundary mesh representation of the phantom that can be used for simulating medical images. An efficient script performs the above actions seamlessly. The user only needs to specify which voxelized breast phantom to use, the compressed thickness, and orientation of the breast.The authors utilized their FE application to simulate compressed states of the breast indicative of mammography and tomosynthesis. Gravity and compression were simulated on example phantoms and used to generate mammograms in the craniocaudal or mediolateral oblique views. The simulated mammograms show a high degree of realism illustrating the utility of the FE method in simulating imaging data of repositioned and compressed breasts.The breast phantoms and the compression software can become a useful resource to the breast imaging research community. These phantoms can then be used to evaluate and compare imaging modalities that involve different positioning and compression of the breast.

Authors
Sturgeon, GM; Kiarashi, N; Lo, JY; Samei, E; Segars, WP
MLA Citation
Sturgeon, GM, Kiarashi, N, Lo, JY, Samei, E, and Segars, WP. "Finite-element modeling of compression and gravity on a population of breast phantoms for multimodality imaging simulation." Medical physics 43.5 (May 2016): 2207-.
PMID
27147333
Source
epmc
Published In
Medical physics
Volume
43
Issue
5
Publish Date
2016
Start Page
2207
DOI
10.1118/1.4945275

A quantitative metrology for performance characterization of five breast tomosynthesis systems based on an anthropomorphic phantom.

In medical imaging systems, proper rendition of anatomy is essential in discerning normal tissue from disease. Currently, digital breast tomosynthesis (DBT) systems are evaluated using subjective evaluation of lesion visibility in uniform phantoms. This study involved the development of a new methodology to objectively measure the rendition of a 3D breast model by an anthropomorphic breast phantom, and its implementation on five clinical DBT systems of different makes and models.A 3D, patient-based breast phantom was fabricated based on XCAT breast models. This phantom was imaged on representative breast tomosynthesis systems. The ability of tomosynthesis systems to accurately reproduce the 3D structure of the breast was assessed by computational analysis of the resultant images in terms of three groups of indices: contrast index (CI), reflective of local difference between adipose and glandular material; adipose variability index (AVI), reflective of contributions of noise and artifacts within uniform adipose regions; and contrast detectability, which describes contrast against local background variability and is described by contrast variability index (CVI), coefficient of variation (COV), contrast to adipose variability index (CAVI), and contrast to noise ratio index (CNRI). The indices were obtained by comparing the image data to the gold standard 3D distribution of breast tissue in the model. Corresponding indices were measured within variable region of interest (ROI) sizes ranging from 10 to 37 mm. The characterization was performed on five tomosynthesis systems: Fuji Aspire Crystal, GE Essential, Hologic Dimension, IMS Giotto, and Siemens Inspiration, all evaluated at a fixed dose of 1.5 mGy average glandular dose, anonymized in random order from A to E.Results are provided as a function of ROI size. The systems ranked orders in terms of CI with values of 7.4%, 7.0%, 6.9%, 6.4%, and 5.2% for systems A-E, respectively. This system ranking was identical for CNRI. Both CI and CNRI were constant over ROI size. The ranking was similar for CVI. The COV also changed little with ROI size and was similar across systems. For 10 mm ROIs, the average system COV was 0.7, which reduced to 0.5 with 37 mm ROIs. Two systems (A and B) exhibited highest AVI values when measured in 10 mm ROIs. This, however, was ROI-size-dependent with the three other systems (C-E) yielding higher AVI values when measured with 37 mm ROIs. Two systems (B and E) showed inferior CAVI compared to others.The quality of rendition tracked with differences in image appearance across systems. The findings illustrate that the anthropomorphic phantom can be used as a basis to extract quantitative values of image attributes in DBT.

Authors
Ikejimba, L; Lo, JY; Chen, Y; Oberhofer, N; Kiarashi, N; Samei, E
MLA Citation
Ikejimba, L, Lo, JY, Chen, Y, Oberhofer, N, Kiarashi, N, and Samei, E. "A quantitative metrology for performance characterization of five breast tomosynthesis systems based on an anthropomorphic phantom." Medical physics 43.4 (April 2016): 1627-.
PMID
27036562
Source
epmc
Published In
Medical physics
Volume
43
Issue
4
Publish Date
2016
Start Page
1627
DOI
10.1118/1.4943373

Radiology Trainee Performance in Digital Breast Tomosynthesis: Relationship Between Difficulty and Error-Making Patterns.

The aim of this study was to better understand the relationship between digital breast tomosynthesis (DBT) difficulty and radiology trainee performance.Twenty-seven radiology residents and fellows and three expert breast imagers reviewed 60 DBT studies consisting of unilateral craniocaudal and medial lateral oblique views. Trainees had no prior DBT experience. All readers provided difficulty ratings and final BI-RADS(®) scores. Expert breast imager consensus interpretations were used to determine the ground truth. Trainee sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were calculated for low- and high-difficulty subsets of cases as assessed by each trainee him or herself (self-assessed difficulty) and consensus expert-assessed difficulty.For self-assessed difficulty, the trainee AUC was 0.696 for high-difficulty and 0.704 for low-difficulty cases (P = .753). Trainee sensitivity was 0.776 for high-difficulty and 0.538 for low-difficulty cases (P < .001). Trainee specificity was 0.558 for high-difficulty and 0.810 for low-difficulty cases (P < .001). For expert-assessed difficulty, the trainee AUC was 0.645 for high-difficulty and 0.816 for low-difficulty cases (P < .001). Trainee sensitivity was 0.612 for high-difficulty and .784 for low-difficulty cases (P < .001). Trainee specificity was 0.654 for high-difficulty and 0.765 for low-difficulty cases (P = .021).Cases deemed difficult by experts were associated with decreases in trainee AUC, sensitivity, and specificity. In contrast, for self-assessed more difficult cases, the trainee AUC was unchanged because of increased sensitivity and compensatory decreased specificity. Educators should incorporate these findings when developing educational materials to teach interpretation of DBT.

Authors
Grimm, LJ; Zhang, J; Lo, JY; Johnson, KS; Ghate, SV; Walsh, R; Mazurowski, MA
MLA Citation
Grimm, LJ, Zhang, J, Lo, JY, Johnson, KS, Ghate, SV, Walsh, R, and Mazurowski, MA. "Radiology Trainee Performance in Digital Breast Tomosynthesis: Relationship Between Difficulty and Error-Making Patterns." Journal of the American College of Radiology : JACR 13.2 (February 2016): 198-202.
PMID
26577878
Source
epmc
Published In
Journal of the American College of Radiology
Volume
13
Issue
2
Publish Date
2016
Start Page
198
End Page
202
DOI
10.1016/j.jacr.2015.09.025

Identification of error making patterns in lesion detection on digital breast tomosynthesis using computer-extracted image features

© 2016 SPIE.Digital breast tomosynthesis (DBT) can improve lesion visibility by eliminating the issue of overlapping breast tissue present in mammography. However, this new modality likely requires new approaches to training. The issue of training in DBT is not well explored. We propose a computer-aided educational approach for DBT training. Our hypothesis is that the trainees' educational outcomes will improve if they are presented with cases individually selected to address their weaknesses. In this study, we focus on the question of how to select such cases. Specifically, we propose an algorithm that based on previously acquired reading data predicts which lesions will be missed by the trainee for future cases (i.e., we focus on false negative error). A logistic regression classifier was used to predict the likelihood of trainee error and computer-extracted features were used as the predictors. Reader data from 3 expert breast imagers was used to establish the ground truth and reader data from 5 radiology trainees was used to evaluate the algorithm performance with repeated holdout cross validation. Receiver operating characteristic (ROC) analysis was applied to measure the performance of the proposed individual trainee models. The preliminary experimental results for 5 trainees showed the individual trainee models were able to distinguish the lesions that would be detected from those that would be missed with the average area under the ROC curve of 0.639 (95% CI, 0.580-0.698). The proposed algorithm can be used to identify difficult cases for individual trainees.

Authors
Wang, M; Zhang, J; Grimm, LJ; Ghate, SV; Walsh, R; Johnson, KS; Lo, JY; Mazurowski, MA
MLA Citation
Wang, M, Zhang, J, Grimm, LJ, Ghate, SV, Walsh, R, Johnson, KS, Lo, JY, and Mazurowski, MA. "Identification of error making patterns in lesion detection on digital breast tomosynthesis using computer-extracted image features." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9787
Publish Date
2016
DOI
10.1117/12.2201061

Design, fabrication, and implementation of voxel-based 3D printed textured phantoms for task-based image quality assessment in CT

© 2016 SPIE.In x-ray computed tomography (CT), task-based image quality studies are typically performed using uniform background phantoms with low-contrast signals. Such studies may have limited clinical relevancy for modern non-linear CT systems due to possible influence of background texture on image quality. The purpose of this study was to design and implement anatomically informed textured phantoms for task-based assessment of low-contrast detection. Liver volumes were segmented from 23 abdominal CT cases. The volumes were characterized in terms of texture features from gray-level co-occurrence and run-length matrices. Using a 3D clustered lumpy background (CLB) model, a fitting technique based on a genetic optimization algorithm was used to find the CLB parameters that were most reflective of the liver textures, accounting for CT system factors of spatial blurring and noise. With the modeled background texture as a guide, a cylinder phantom (165 mm in diameter and 30 mm height) was designed, containing 20 low-contrast spherical signals (6 mm in diameter at targeted contrast levels of ∼3.2, 5.2, 7.2, 10, and 14 HU, 4 repeats per signal). The phantom was voxelized and input into a commercial multi-material 3D printer (Object Connex 350), with custom software for voxel-based printing. Using principles of digital half-toning and dithering, the 3D printer was programmed to distribute two base materials (VeroWhite and TangoPlus, nominal voxel size of 42x84x30 microns) to achieve the targeted spatial distribution of x-ray attenuation properties. The phantom was used for task-based image quality assessment of a clinically available iterative reconstruction algorithm (Sinogram Affirmed Iterative Reconstruction, SAFIRE) using a channelized Hotelling observer paradigm. Images of the textured phantom and a corresponding uniform phantom were acquired at six dose levels and observer model performance was estimated for each condition (5 contrasts x 6 doses x 2 reconstructions x 2 backgrounds = 120 total conditions). Based on the observer model results, the dose reduction potential of SAFIRE was computed and compared between the uniform and textured phantom. The dose reduction potential of SAFIRE was found to be 23% based on the uniform phantom and 17% based on the textured phantom. This discrepancy demonstrates the need to consider background texture when assessing non-linear reconstruction algorithms.

Authors
Solomon, J; Ba, A; Diao, A; Lo, J; Bier, E; Bochud, F; Gehm, M; Samei, E
MLA Citation
Solomon, J, Ba, A, Diao, A, Lo, J, Bier, E, Bochud, F, Gehm, M, and Samei, E. "Design, fabrication, and implementation of voxel-based 3D printed textured phantoms for task-based image quality assessment in CT." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9783
Publish Date
2016
DOI
10.1117/12.2217463

Second generation anthropomorphic physical phantom for mammography and DBT: Incorporating voxelized 3D printing and inkjet printing of iodinated lesion inserts

© 2016 SPIE.Physical phantoms are needed for the evaluation and optimization of new digital breast tomosynthesis (DBT) systems. Previously, we developed an anthropomorphic phantom based on human subject breast CT data and fabricated using commercial 3D printing. We now present three key advancements: voxelized 3D printing, photopolymer material doping, and 2D inkjet printing of lesion inserts. First, we bypassed the printer's control software in order to print in voxelized form instead of conventional STL surfaces, thus improving resolution and allowing dithering to mix the two photopolymer materials into arbitrary proportions. We demonstrated ability to print details as small as 150μm, and dithering to combine VeroWhitePlus and TangoPlus in 10% increments. Second, to address the limited attenuation difference among commercial photopolymers, we evaluated a beta sample from Stratasys with increased TiO2 doping concentration up to 2.5%, which corresponded to 98% breast density. By spanning 36% to 98% breast density, this doubles our previous contrast. Third, using inkjet printers modified to print with iopamidol, we created 2D lesion patterns on paper that can be sandwiched into the phantom. Inkjet printing has advantages of being inexpensive and easy, and more contrast can be delivered through overprinting. Printing resolution was maintained at 210 μm horizontally and 330 μm vertically even after 10 overprints. Contrast increased linearly with overprinting at 0.7% per overprint. Together, these three new features provide the basis for creating a new anthropomorphic physical breast phantom with improved resolution and contrast, as well as the ability to insert 2D lesions for task-based assessment of performance.

Authors
Sikaria, D; Musinsky, S; Sturgeon, GM; Solomon, J; Diao, A; Gehm, ME; Samei, E; Glick, SJ; Lo, JY
MLA Citation
Sikaria, D, Musinsky, S, Sturgeon, GM, Solomon, J, Diao, A, Gehm, ME, Samei, E, Glick, SJ, and Lo, JY. "Second generation anthropomorphic physical phantom for mammography and DBT: Incorporating voxelized 3D printing and inkjet printing of iodinated lesion inserts." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9783
Publish Date
2016
DOI
10.1117/12.2217667

Eigenbreasts for statistical breast phantoms

© 2016 SPIE.To facilitate rigorous virtual clinical trials using model observers for breast imaging optimization and evaluation, we demonstrated a method of defining statistical models, based on 177 sets of breast CT patient data, in order to generate tens of thousands of unique digital breast phantoms. In order to separate anatomical texture from variation in breast shape, each training set of breast phantoms were deformed to a consistent atlas compressed geometry. Principal component analysis (PCA) was then performed on the shape-matched breast CT volumes to capture the variation of patient breast textures. PCA decomposes the training set of N breast CT volumes into an N-1-dimensional space of eigenvectors, which we call eigenbreasts. By summing weighted combinations of eigenbreasts, a large ensemble of different breast phantoms can be newly created. Different training sets can be used in eigenbreast analysis for designing basis models to target sub-populations defined by breast characteristics, such as size or density. In this work, we plan to generate ensembles of 30,000 new phantoms based on glandularity for an upcoming virtual trial of lesion detectability in digital breast tomosynthesis. Our method extends our series of digital and physical breast phantoms based on human subject anatomy, providing the capability to generate new, unique ensembles consisting of tens of thousands or more virtual subjects. This work represents an important step towards conducting future virtual trials for tasks-based assessment of breast imaging, where it is vital to have a large ensemble of realistic phantoms for statistical power as well as clinical relevance.

Authors
Sturgeon, GM; Tward, DJ; Ketcha, M; Ratnanather, JT; Miller, MI; Park, S; Segars, WP; Lo, JY
MLA Citation
Sturgeon, GM, Tward, DJ, Ketcha, M, Ratnanather, JT, Miller, MI, Park, S, Segars, WP, and Lo, JY. "Eigenbreasts for statistical breast phantoms." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9783
Publish Date
2016
DOI
10.1117/12.2216398

Comparison of model and human observer performance in FFDM, DBT, and synthetic mammography

© 2016 SPIE.Reader studies are important in assessing breast imaging systems. The purpose of this work was to assess task-based performance of full field digital mammography (FFDM), digital breast tomosynthesis (DBT), and synthetic mammography (SM) using different phantom types, and to determine an accurate observer model for human readers. Images were acquired on a Hologic Selenia Dimensions system with a uniform and anthropomorphic phantom. A contrast detail insert of small, low-contrast disks was created using an inkjet printer with iodine-doped ink and inserted in the phantoms. The disks varied in diameter from 210 to 630 μm, and in contrast from 1.1% contrast to 2.2% in regular increments. Human and model observers performed a 4-alternative forced choice experiment. The models were a non-prewhitening matched filter with eye model (NPWE) and a channelized Hotelling observer with either Gabor channels (Gabor-CHO) or Laguerre-Gauss channels (LG-CHO). With the given phantoms, reader scores were higher in FFDM and DBT than SM. The structure in the phantom background had a bigger impact on outcome for DBT than for FFDM or SM. All three model observers showed good correlation with humans in the uniform background, with ρ between 0.89 and 0.93. However, in the structured background, only the CHOs had high correlation, with ρ=0.92 for Gabor-CHO, 0.90 for LG-CHO, and 0.77 for NPWE. Because results of any analysis can depend on the phantom structure, conclusions of modality performance may need to be taken in the context of an appropriate model observer and a realistic phantom.

Authors
Ikejimba, L; Glick, SJ; Samei, E; Lo, JY
MLA Citation
Ikejimba, L, Glick, SJ, Samei, E, and Lo, JY. "Comparison of model and human observer performance in FFDM, DBT, and synthetic mammography." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9783
Publish Date
2016
DOI
10.1117/12.2216858

Investigation of optimal parameters for penalized maximum-likelihood reconstruction applied to iodinated contrast-enhanced breast CT

© 2016 SPIE.Although digital mammography has reduced breast cancer mortality by approximately 30%, sensitivity and specificity are still far from perfect. In particular, the performance of mammography is especially limited for women with dense breast tissue. Two out of every three biopsies performed in the U.S. are unnecessary, thereby resulting in increased patient anxiety, pain, and possible complications. One promising tomographic breast imaging method that has recently been approved by the FDA is dedicated breast computed tomography (BCT). However, visualizing lesions with BCT can still be challenging for women with dense breast tissue due to the minimal contrast for lesions surrounded by fibroglandular tissue. In recent years there has been renewed interest in improving lesion conspicuity in x-ray breast imaging by administration of an iodinated contrast agent. Due to the fully 3-D imaging nature of BCT, as well as sub-optimal contrast enhancement while the breast is under compression with mammography and breast tomosynthesis, dedicated BCT of the uncompressed breast is likely to offer the best solution for injected contrast-enhanced x-ray breast imaging. It is well known that use of statistically-based iterative reconstruction in CT results in improved image quality at lower radiation dose. Here we investigate possible improvements in image reconstruction for BCT, by optimizing free regularization parameter in method of maximum likelihood and comparing its performance with clinical cone-beam filtered backprojection (FBP) algorithm.

Authors
Makeev, A; Ikejimba, L; Lo, JY; Glick, SJ
MLA Citation
Makeev, A, Ikejimba, L, Lo, JY, and Glick, SJ. "Investigation of optimal parameters for penalized maximum-likelihood reconstruction applied to iodinated contrast-enhanced breast CT." January 1, 2016.
Source
scopus
Published In
Proceedings of SPIE
Volume
9783
Publish Date
2016
DOI
10.1117/12.2217438

Population of 224 realistic human subject-based computational breast phantoms.

To create a database of highly realistic and anatomically variable 3D virtual breast phantoms based on dedicated breast computed tomography (bCT) data.A tissue classification and segmentation algorithm was used to create realistic and detailed 3D computational breast phantoms based on 230 + dedicated bCT datasets from normal human subjects. The breast volume was identified using a coarse three-class fuzzy C-means segmentation algorithm which accounted for and removed motion blur at the breast periphery. Noise in the bCT data was reduced through application of a postreconstruction 3D bilateral filter. A 3D adipose nonuniformity (bias field) correction was then applied followed by glandular segmentation using a 3D bias-corrected fuzzy C-means algorithm. Multiple tissue classes were defined including skin, adipose, and several fractional glandular densities. Following segmentation, a skin mask was produced which preserved the interdigitated skin, adipose, and glandular boundaries of the skin interior. Finally, surface modeling was used to produce digital phantoms with methods complementary to the XCAT suite of digital human phantoms.After rejecting some datasets due to artifacts, 224 virtual breast phantoms were created which emulate the complex breast parenchyma of actual human subjects. The volume breast density (with skin) ranged from 5.5% to 66.3% with a mean value of 25.3% ± 13.2%. Breast volumes ranged from 25.0 to 2099.6 ml with a mean value of 716.3 ± 386.5 ml. Three breast phantoms were selected for imaging with digital compression (using finite element modeling) and simple ray-tracing, and the results show promise in their potential to produce realistic simulated mammograms.This work provides a new population of 224 breast phantoms based on in vivo bCT data for imaging research. Compared to previous studies based on only a few prototype cases, this dataset provides a rich source of new cases spanning a wide range of breast types, volumes, densities, and parenchymal patterns.

Authors
Erickson, DW; Wells, JR; Sturgeon, GM; Samei, E; Dobbins, JT; Segars, WP; Lo, JY
MLA Citation
Erickson, DW, Wells, JR, Sturgeon, GM, Samei, E, Dobbins, JT, Segars, WP, and Lo, JY. "Population of 224 realistic human subject-based computational breast phantoms." Medical physics 43.1 (January 2016): 23-.
PMID
26745896
Source
epmc
Published In
Medical physics
Volume
43
Issue
1
Publish Date
2016
Start Page
23
DOI
10.1118/1.4937597

Semiautomated head-and-neck IMRT planning using dose warping and scaling to robustly adapt plans in a knowledge database containing potentially suboptimal plans.

Prior work by the authors and other groups has studied the creation of automated intensity modulated radiotherapy (IMRT) plans of equivalent quality to those in a patient database of manually created clinical plans; those database plans provided guidance on the achievable sparing to organs-at-risk (OARs). However, in certain sites, such as head-and-neck, the clinical plans may not be sufficiently optimized because of anatomical complexity and clinical time constraints. This could lead to automated plans that suboptimally exploit OAR sparing. This work investigates a novel dose warping and scaling scheme that attempts to reduce effects of suboptimal sparing in clinical database plans, thus improving the quality of semiautomated head-and-neck cancer (HNC) plans.Knowledge-based radiotherapy (KBRT) plans for each of ten "query" patients were semiautomatically generated by identifying the most similar "match" patient in a database of 103 clinical manually created patient plans. The match patient's plans were adapted to the query case by: (1) deforming the match beam fluences to suit the query target volume and (2) warping the match primary/boost dose distribution to suit the query geometry and using the warped distribution to generate query primary/boost optimization dose-volume constraints. Item (2) included a distance scaling factor to improve query OAR dose sparing with respect to the possibly suboptimal clinical match plan. To further compensate for a component plan of the match case (primary/boost) not optimally sparing OARs, the query dose volume constraints were reduced using a dose scaling factor to be the minimum from either (a) the warped component plan (primary or boost) dose distribution or (b) the warped total plan dose distribution (primary + boost) scaled in proportion to the ratio of component prescription dose to total prescription dose. The dose-volume constraints were used to plan the query case with no human intervention to adjust constraints during plan optimization.KBRT and original clinical plans were dosimetrically equivalent for parotid glands (mean/median doses), spinal cord, and brainstem (maximum doses). KBRT plans significantly reduced larynx median doses (21.5 ± 6.6 Gy to 17.9 ± 3.9 Gy), and oral cavity mean (32.3 ± 6.2 Gy to 28.9 ± 5.4 Gy) and median (28.7 ± 5.7 Gy to 23.2 ± 5.3 Gy) doses. Doses to ipsilateral parotid gland, larynx, oral cavity, and brainstem were lower or equivalent in the KBRT plans for the majority of cases. By contrast, KBRT plans generated without the dose warping and dose scaling steps were not significantly different from the clinical plans.Fast, semiautomatically generated HNC IMRT plans adapted from existing plans in a clinical database can be of equivalent or better quality than manually created plans. The reductions in OAR doses in the semiautomated plans, compared to the clinical plans, indicate that the proposed dose warping and scaling method shows promise in mitigating the impact of suboptimal clinical plans.

Authors
Schmidt, M; Lo, JY; Grzetic, S; Lutzky, C; Brizel, DM; Das, SK
MLA Citation
Schmidt, M, Lo, JY, Grzetic, S, Lutzky, C, Brizel, DM, and Das, SK. "Semiautomated head-and-neck IMRT planning using dose warping and scaling to robustly adapt plans in a knowledge database containing potentially suboptimal plans." Medical physics 42.8 (August 2015): 4428-4434.
PMID
26233173
Source
epmc
Published In
Medical physics
Volume
42
Issue
8
Publish Date
2015
Start Page
4428
End Page
4434
DOI
10.1118/1.4923174

Development of realistic physical breast phantoms matched to virtual breast phantoms based on human subject data.

Physical phantoms are essential for the development, optimization, and evaluation of x-ray breast imaging systems. Recognizing the major effect of anatomy on image quality and clinical performance, such phantoms should ideally reflect the three-dimensional structure of the human breast. Currently, there is no commercially available three-dimensional physical breast phantom that is anthropomorphic. The authors present the development of a new suite of physical breast phantoms based on human data.The phantoms were designed to match the extended cardiac-torso virtual breast phantoms that were based on dedicated breast computed tomography images of human subjects. The phantoms were fabricated by high-resolution multimaterial additive manufacturing (3D printing) technology. The glandular equivalency of the photopolymer materials was measured relative to breast tissue-equivalent plastic materials. Based on the current state-of-the-art in the technology and available materials, two variations were fabricated. The first was a dual-material phantom, the Doublet. Fibroglandular tissue and skin were represented by the most radiographically dense material available; adipose tissue was represented by the least radiographically dense material. The second variation, the Singlet, was fabricated with a single material to represent fibroglandular tissue and skin. It was subsequently filled with adipose-equivalent materials including oil, beeswax, and permanent urethane-based polymer. Simulated microcalcification clusters were further included in the phantoms via crushed eggshells. The phantoms were imaged and characterized visually and quantitatively.The mammographic projections and tomosynthesis reconstructed images of the fabricated phantoms yielded realistic breast background. The mammograms of the phantoms demonstrated close correlation with simulated mammographic projection images of the corresponding virtual phantoms. Furthermore, power-law descriptions of the phantom images were in general agreement with real human images. The Singlet approach offered more realistic contrast as compared to the Doublet approach, but at the expense of air bubbles and air pockets that formed during the filling process.The presented physical breast phantoms and their matching virtual breast phantoms offer realistic breast anatomy, patient variability, and ease of use, making them a potential candidate for performing both system quality control testing and virtual clinical trials.

Authors
Kiarashi, N; Nolte, AC; Sturgeon, GM; Segars, WP; Ghate, SV; Nolte, LW; Samei, E; Lo, JY
MLA Citation
Kiarashi, N, Nolte, AC, Sturgeon, GM, Segars, WP, Ghate, SV, Nolte, LW, Samei, E, and Lo, JY. "Development of realistic physical breast phantoms matched to virtual breast phantoms based on human subject data." Medical physics 42.7 (July 2015): 4116-4126.
PMID
26133612
Source
epmc
Published In
Medical physics
Volume
42
Issue
7
Publish Date
2015
Start Page
4116
End Page
4126
DOI
10.1118/1.4919771

Does Breast Imaging Experience During Residency Translate Into Improved Initial Performance in Digital Breast Tomosynthesis?

To determine the initial digital breast tomosynthesis (DBT) performance of radiology trainees with varying degrees of breast imaging experience.To test trainee performance with DBT, we performed a reader study, after obtaining IRB approval. Two medical students, 20 radiology residents, 4 nonbreast imaging fellows, 3 breast imaging fellows, and 3 fellowship-trained breast imagers reviewed 60 unilateral DBT studies (craniocaudal and medio-lateral oblique views). Trainees had no DBT experience. Each reader recorded a final BI-RADS assessment for each case. The consensus interpretations from fellowship-trained breast imagers were used to establish the ground truth. Area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were calculated. For analysis, first- through third-year residents were classified as junior trainees, and fourth-year residents plus nonbreast imaging fellows were classified as senior trainees.The AUCs were .569 for medical students, .721 for junior trainees, .701 for senior trainees, and .792 for breast imaging fellows. The junior and senior trainee AUCs were equivalent (P < .01) using a two one-sided test for equivalence, with a significance threshold of 0.1. The sensitivities and specificities were highest for breast imaging fellows (.778 and .815 respectively), but similar for junior (.631 and .714, respectively) and senior trainees (.678 and .661, respectively).Initial performance with DBT among radiology residents and nonbreast imaging fellows is independent of years of training. Radiology educators should consider these findings when developing educational materials.

Authors
Zhang, J; Grimm, LJ; Lo, JY; Johnson, KS; Ghate, SV; Walsh, R; Mazurowski, MA
MLA Citation
Zhang, J, Grimm, LJ, Lo, JY, Johnson, KS, Ghate, SV, Walsh, R, and Mazurowski, MA. "Does Breast Imaging Experience During Residency Translate Into Improved Initial Performance in Digital Breast Tomosynthesis?." Journal of the American College of Radiology : JACR 12.7 (July 2015): 728-732.
PMID
26143567
Source
epmc
Published In
Journal of the American College of Radiology
Volume
12
Issue
7
Publish Date
2015
Start Page
728
End Page
732
DOI
10.1016/j.jacr.2015.02.025

Incorporating breast tomosynthesis into radiology residency: Does trainee experience in breast imaging translate into improved performance with this new modality?

© 2015 SPIE.Digital breast tomosynthesis (DBT) is a powerful new imaging modality that has the potential to transform breast cancer screening practices. The advantages over mammography include improved sensitivity and specificity as well as the detection of additional invasive cancers. While this modality holds many advantages, the best means of incorporating DBT into radiology training programs is currently not well understood. The initial performance of a trainee in DBT might depend on the amount of previous radiology training, in particular breast imaging experience. In our study, we tested the DBT interpretive skills of radiology trainees with different levels of breast imaging training, but with no prior DBT experience. We recruited 16 radiology trainees to review 60 DBT studies. A fellowship-trained expert breast radiologist reviewed the studies and provided the gold standard interpretations. Receiver operating characteristic analysis was used to evaluate the performance of the trainees. Our results show that there is no notable difference in trainee performance in DBT, regardless of the number of years of general radiology experience. These results provide guidance to breast imaging educators as they prepare new curricula to teach DBT. These curricula should provide a base training which will likely be suitable for all trainee levels.

Authors
Grimm, LJ; Zhang, J; Johnson, KS; Lo, JY; Mazurowski, MA
MLA Citation
Grimm, LJ, Zhang, J, Johnson, KS, Lo, JY, and Mazurowski, MA. "Incorporating breast tomosynthesis into radiology residency: Does trainee experience in breast imaging translate into improved performance with this new modality?." January 1, 2015.
Source
scopus
Published In
Proceedings of SPIE
Volume
9416
Publish Date
2015
DOI
10.1117/12.2082810

The impact of breast structure on lesion detection in breast tomosynthesis

© 2015 SPIE.Virtual clinical trials (VCT) can be carefully designed to inform, orient, or potentially replace clinical trials. The focus of this study was to demonstrate the capability of the sophisticated tools that can be used in the design, implementation, and performance analysis of VCTs, through characterization of the effect of background tissue density and heterogeneity on the detection of irregular masses in digital breast tomosynthesis. Twenty breast phantoms from the extended cardiactorso (XCAT) family, generated based on dedicated breast computed tomography of human subjects, were used to extract a total of 2173 volumes of interest (VOI) from simulated tomosynthesis images. Five different lesions, modeled after human subject tomosynthesis images, were embedded in the breasts, for a total of 6×2173 VOIs with and without lesions. Effects of background tissue density and heterogeneity on the detection of the lesions were studied by implementing a doubly composite hypothesis signal detection theory paradigm with location known exactly, lesion known exactly, and background known statistically. The results indicated that the detection performance as measured by the area under the receiver operating characteristic curve (ROC) deteriorated as density was increased, yielding findings consistent with clinical studies. The detection performance varied substantially across the twenty breasts. Furthermore, the log-likelihood ratio under H0 and H1seemed to be affected by background tissue density and heterogeneity differently. Considering background tissue variability can change the outcomes of a VCT and is hence of crucial importance. The XCAT breast phantoms can address this concern by offering realistic modeling of background tissue variability based on a wide range of human subjects.

Authors
Kiarashi, N; Nolte, LW; Lo, JY; Segars, WP; Ghate, SV; Samei, E
MLA Citation
Kiarashi, N, Nolte, LW, Lo, JY, Segars, WP, Ghate, SV, and Samei, E. "The impact of breast structure on lesion detection in breast tomosynthesis." January 1, 2015.
Source
scopus
Published In
Proceedings of SPIE
Volume
9412
Publish Date
2015
DOI
10.1117/12.2082473

A quantitative metrology for performance characterization of breast tomosynthesis systems based on an anthropomorphic phantom

© 2015 SPIE.Purpose: Common methods for assessing image quality of digital breast tomosynthesis (DBT) devices currently utilize simplified or otherwise unrealistic phantoms, which use inserts in a uniform background and gauge performance based on a subjective evaluation of insert visibility. This study proposes a different methodology to assess system performance using a three-dimensional clinically-informed anthropomorphic breast phantom. Methods: The system performance is assessed by imaging the phantom and computationally characterizing the resultant images in terms of several new metrics. These include a contrast index (reflective of local difference between adipose and glandular material), a contrast to noise ratio index (reflective of contrast against local background noise), and a nonuniformity index (reflective of contributions of noise and artifacts within uniform adipose regions). Indices were measured at ROI sizes of 10mm and 37 mm, respectively. The method was evaluated at fixed dose of 1.5 mGy AGD. Results: Results indicated notable differences between systems. At 10 mm, vendor A had the highest contrast index, followed by B and C in that. The performance ranking was identical at the largest ROI size. The non-uniformity index similarly exhibited system-dependencies correlated with visual appearance of clutter from out-of-plane artifacts. Vendor A had the greatest NI at all ROI sizes, B had the second greatest, and C the least. Conclusions: The findings illustrate that the anthropomorphic phantom can be used as a quality control tool with results that are targeted to be more reflective of clinical performance of breast tomosynthesis systems of multiple manufacturers.

Authors
Ikejimba, L; Chen, Y; Oberhofer, N; Kiarashi, N; Lo, JY; Samei, E
MLA Citation
Ikejimba, L, Chen, Y, Oberhofer, N, Kiarashi, N, Lo, JY, and Samei, E. "A quantitative metrology for performance characterization of breast tomosynthesis systems based on an anthropomorphic phantom." January 1, 2015.
Source
scopus
Published In
Proceedings of SPIE
Volume
9412
Publish Date
2015
DOI
10.1117/12.2082594

Using computer-extracted image features for modeling of error-making patterns in detection of mammographic masses among radiology residents.

Mammography is the most widely accepted and utilized screening modality for early breast cancer detection. Providing high quality mammography education to radiology trainees is essential, since excellent interpretation skills are needed to ensure the highest benefit of screening mammography for patients. The authors have previously proposed a computer-aided education system based on trainee models. Those models relate human-assessed image characteristics to trainee error. In this study, the authors propose to build trainee models that utilize features automatically extracted from images using computer vision algorithms to predict likelihood of missing each mass by the trainee. This computer vision-based approach to trainee modeling will allow for automatically searching large databases of mammograms in order to identify challenging cases for each trainee.The authors' algorithm for predicting the likelihood of missing a mass consists of three steps. First, a mammogram is segmented into air, pectoral muscle, fatty tissue, dense tissue, and mass using automated segmentation algorithms. Second, 43 features are extracted using computer vision algorithms for each abnormality identified by experts. Third, error-making models (classifiers) are applied to predict the likelihood of trainees missing the abnormality based on the extracted features. The models are developed individually for each trainee using his/her previous reading data. The authors evaluated the predictive performance of the proposed algorithm using data from a reader study in which 10 subjects (7 residents and 3 novices) and 3 experts read 100 mammographic cases. Receiver operating characteristic (ROC) methodology was applied for the evaluation.The average area under the ROC curve (AUC) of the error-making models for the task of predicting which masses will be detected and which will be missed was 0.607 (95% CI,0.564-0.650). This value was statistically significantly different from 0.5 (p<0.0001). For the 7 residents only, the AUC performance of the models was 0.590 (95% CI,0.537-0.642) and was also significantly higher than 0.5 (p=0.0009). Therefore, generally the authors' models were able to predict which masses were detected and which were missed better than chance.The authors proposed an algorithm that was able to predict which masses will be detected and which will be missed by each individual trainee. This confirms existence of error-making patterns in the detection of masses among radiology trainees. Furthermore, the proposed methodology will allow for the optimized selection of difficult cases for the trainees in an automatic and efficient manner.

Authors
Zhang, J; Lo, JY; Kuzmiak, CM; Ghate, SV; Yoon, SC; Mazurowski, MA
MLA Citation
Zhang, J, Lo, JY, Kuzmiak, CM, Ghate, SV, Yoon, SC, and Mazurowski, MA. "Using computer-extracted image features for modeling of error-making patterns in detection of mammographic masses among radiology residents." Medical physics 41.9 (September 2014): 091907-.
PMID
25186394
Source
epmc
Published In
Medical physics
Volume
41
Issue
9
Publish Date
2014
Start Page
091907
DOI
10.1118/1.4892173

Radiation dosimetry in digital breast tomosynthesis: report of AAPM Tomosynthesis Subcommittee Task Group 223.

The radiation dose involved in any medical imaging modality that uses ionizing radiation needs to be well understood by the medical physics and clinical community. This is especially true of screening modalities. Digital breast tomosynthesis (DBT) has recently been introduced into the clinic and is being used for screening for breast cancer in the general population. Therefore, it is important that the medical physics community have the required information to be able to understand, estimate, and communicate the radiation dose levels involved in breast tomosynthesis imaging. For this purpose, the American Association of Physicists in Medicine Task Group 223 on Dosimetry in Tomosynthesis Imaging has prepared this report that discusses dosimetry in breast imaging in general, and describes a methodology and provides the data necessary to estimate mean breast glandular dose from a tomosynthesis acquisition. In an effort to maximize familiarity with the procedures and data provided in this Report, the methodology to perform the dose estimation in DBT is based as much as possible on that used in mammography dose estimation.

Authors
Sechopoulos, I; Sabol, JM; Berglund, J; Bolch, WE; Brateman, L; Christodoulou, E; Flynn, M; Geiser, W; Goodsitt, M; Jones, AK; Lo, JY; Maidment, ADA; Nishino, K; Nosratieh, A; Ren, B; Segars, WP; Von Tiedemann, M
MLA Citation
Sechopoulos, I, Sabol, JM, Berglund, J, Bolch, WE, Brateman, L, Christodoulou, E, Flynn, M, Geiser, W, Goodsitt, M, Jones, AK, Lo, JY, Maidment, ADA, Nishino, K, Nosratieh, A, Ren, B, Segars, WP, and Von Tiedemann, M. "Radiation dosimetry in digital breast tomosynthesis: report of AAPM Tomosynthesis Subcommittee Task Group 223." Medical physics 41.9 (September 2014): 091501-.
PMID
25186375
Source
epmc
Published In
Medical physics
Volume
41
Issue
9
Publish Date
2014
Start Page
091501
DOI
10.1118/1.4892600

Development and application of a suite of 4-D virtual breast phantoms for optimization and evaluation of breast imaging systems.

Mammography is currently the most widely utilized tool for detection and diagnosis of breast cancer. However, in women with dense breast tissue, tissue overlap may obscure lesions. Digital breast tomosynthesis can reduce tissue overlap. Furthermore, imaging with contrast enhancement can provide additional functional information about lesions, such as morphology and kinetics, which in turn may improve lesion identification and characterization. The performance of these imaging techniques is strongly dependent on the structural composition of the breast, which varies significantly among patients. Therefore, imaging system and imaging technique optimization should take patient variability into consideration. Furthermore, optimization of imaging techniques that employ contrast agents should include the temporally varying breast composition with respect to the contrast agent uptake kinetics. To these ends, we have developed a suite of 4-D virtual breast phantoms, which are incorporated with the kinetics of contrast agent propagation in different tissues and can realistically model normal breast parenchyma as well as benign and malignant lesions. This development presents a new approach in performing simulation studies using truly anthropomorphic models. To demonstrate the utility of the proposed 4-D phantoms, we present a simplified example study to compare the performance of 14 imaging paradigms qualitatively and quantitatively.

Authors
Kiarashi, N; Lo, JY; Lin, Y; Ikejimba, LC; Ghate, SV; Nolte, LW; Dobbins, JT; Segars, WP; Samei, E
MLA Citation
Kiarashi, N, Lo, JY, Lin, Y, Ikejimba, LC, Ghate, SV, Nolte, LW, Dobbins, JT, Segars, WP, and Samei, E. "Development and application of a suite of 4-D virtual breast phantoms for optimization and evaluation of breast imaging systems." IEEE transactions on medical imaging 33.7 (July 2014): 1401-1409.
PMID
24691118
Source
epmc
Published In
IEEE Transactions on Medical Imaging
Volume
33
Issue
7
Publish Date
2014
Start Page
1401
End Page
1409
DOI
10.1109/tmi.2014.2312733

Task-based strategy for optimized contrast enhanced breast imaging: Analysis of six imaging techniques for mammography and tomosynthesis

Authors
Ikejimba, LC; Kiarashi, N; Ghate, SV; Samei, E; Lo, JY
MLA Citation
Ikejimba, LC, Kiarashi, N, Ghate, SV, Samei, E, and Lo, JY. "Task-based strategy for optimized contrast enhanced breast imaging: Analysis of six imaging techniques for mammography and tomosynthesis." Medical Physics 41.6 (May 20, 2014): 061908-061908.
Source
crossref
Published In
Medical physics
Volume
41
Issue
6
Publish Date
2014
Start Page
061908
End Page
061908
DOI
10.1118/1.4873317

Modeling resident error-making patterns in detection of mammographic masses using computer-extracted image features: Preliminary experiments

Providing high quality mammography education to radiology trainees is essential, as good interpretation skills potentially ensure the highest benefit of screening mammography for patients. We have previously proposed a computer-aided education system that utilizes trainee models, which relate human-assessed image characteristics to interpretation error. We proposed that these models be used to identify the most difficult and therefore the most educationally useful cases for each trainee. In this study, as a next step in our research, we propose to build trainee models that utilize features that are automatically extracted from images using computer vision algorithms. To predict error, we used a logistic regression which accepts imaging features as input and returns error as output. Reader data from 3 experts and 3 trainees were used. Receiver operating characteristic analysis was applied to evaluate the proposed trainee models. Our experiments showed that, for three trainees, our models were able to predict error better than chance. This is an important step in the development of adaptive computer-aided education systems since computer-extracted features will allow for faster and more extensive search of imaging databases in order to identify the most educationally beneficial cases. © 2014 SPIE.

Authors
Mazurowski, MA; Zhang, J; Lo, JY; Kuzmiak, CM; Ghate, SV; Yoon, S
MLA Citation
Mazurowski, MA, Zhang, J, Lo, JY, Kuzmiak, CM, Ghate, SV, and Yoon, S. "Modeling resident error-making patterns in detection of mammographic masses using computer-extracted image features: Preliminary experiments." January 1, 2014.
Source
scopus
Published In
Proceedings of SPIE
Volume
9037
Publish Date
2014
DOI
10.1117/12.2044404

Population of 100 realistic, patient-based computerized breast phantoms for multi-modality imaging research

Breast imaging is an important area of research with many new Techniques being investigated To further reduce The morbidity and mortality of breast cancer Through early detection. Computerized phantoms can provide an essential Tool To quantitatively compare new imaging systems and Techniques. Current phantoms, however, lack sufficient realism in depicting The complex 3D anatomy of The breast. In This work, we created one-hundred realistic and detailed 3D computational breast phantoms based on high-resolution CT datasets from normal patients. We also developed a finiteelement application To simulate different compression states of The breast, making The phantoms applicable To multimodality imaging research. The breast phantoms and Tools developed in This work were packaged into user-friendly software applications To distribute for breast imaging research. © 2014 SPIE.

Authors
Segars, WP; Veress, AI; Wells, JR; Sturgeon, GM; Kiarashi, N; Lo, JY; Samei, E; Dobbins, JT
MLA Citation
Segars, WP, Veress, AI, Wells, JR, Sturgeon, GM, Kiarashi, N, Lo, JY, Samei, E, and Dobbins, JT. "Population of 100 realistic, patient-based computerized breast phantoms for multi-modality imaging research." January 1, 2014.
Source
scopus
Published In
Proceedings of SPIE
Volume
9033
Publish Date
2014
DOI
10.1117/12.2043868

A second generation of physical anthropomorphic 3D breast phantoms based on human subject data

Previous fabrication of anthropomorphic breast phantoms has demonstrated Their viability as a model for 2D (mammography) and 3D (tomosynthesis) breast imaging systems. Further development of These models will be essential for The evaluation of breast x-ray systems. There is also The potential To use Them as The ground Truth in virtual clinical Trials. The first generation of phantoms was segmented from human subject dedicated breast computed Tomography data and fabricated into physical models using highresolution 3D printing. Two variations were made. The first was a multi-material model (doublet) printed with Two photopolymers To represent glandular and adipose Tissues with The greatest physical contrast available, mimicking 75% and 35% glandular Tissue. The second model was printed with a single 75% glandular equivalent photopolymer (singlet) To represent glandular Tissue, which can be filled independently with an adipose-equivalent material such as oil. For This study, we have focused on improving The latter, The singlet phantom. First, The Temporary oil filler has been replaced with a permanent adipose-equivalent urethane-based polymer. This offers more realistic contrast as compared To The multi-material approach at The expense of air bubbles and pockets That form during The filling process. Second, microcalcification clusters have been included in The singlet model via crushed eggshells, which have very similar chemical composition To calcifications in vivo. The results from These new prototypes demonstrate significant improvement over The first generation of anthropomorphic physical phantoms. © 2014 SPIE.

Authors
Nolte, A; Kiarashi, N; Samei, E; Segars, WP; Lo, JY
MLA Citation
Nolte, A, Kiarashi, N, Samei, E, Segars, WP, and Lo, JY. "A second generation of physical anthropomorphic 3D breast phantoms based on human subject data." January 1, 2014.
Source
scopus
Published In
Proceedings of SPIE
Volume
9033
Publish Date
2014
DOI
10.1117/12.2043703

A task-based comparison of two reconstruction algorithms for digital breast tomosynthesis

Digital breast tomosynthesis (DBT) generates 3-D reconstructions of the breast by taking X-Ray projections at various angles around the breast. DBT improves cancer detection as it minimizes tissue overlap that is present in traditional 2-D mammography. In this work, two methods of reconstruction, filtered backprojection (FBP) and the Newton-Raphson iterative reconstruction were used to create 3-D reconstructions from phantom images acquired on a breast tomosynthesis system. The task based image analysis method was used to compare the performance of each reconstruction technique. The task simulated a 10mm lesion within the breast containing iodine concentrations between 0.0mg/ml and 8.6mg/ml. The TTF was calculated using the reconstruction of an edge phantom, and the NPS was measured with a structured breast phantom (CIRS 020) over different exposure levels. The detectability index d’was calculated to assess image quality of the reconstructed phantom images. Image quality was assessed for both conventional, single energy and dual energy subtracted reconstructions. Dose allocation between the high and low energy scans was also examined. Over the full range of dose allocations, the iterative reconstruction yielded a higher detectability index than the FBP for single energy reconstructions. For dual energy subtraction, detectability index was maximized when most of the dose was allocated to the high energy image. With that dose allocation, the performance trend for reconstruction algorithms reversed; FBP performed better than the corresponding iterative reconstruction. However, FBP performance varied very erratically with changing dose allocation. Therefore, iterative reconstruction is preferred for both imaging modalities despite underperforming dual energy FBP, as it provides stable results. © 2014 SPIE.

Authors
Mahadevan, R; Ikejimba, LC; Lin, Y; Samei, E; Lo, JY
MLA Citation
Mahadevan, R, Ikejimba, LC, Lin, Y, Samei, E, and Lo, JY. "A task-based comparison of two reconstruction algorithms for digital breast tomosynthesis." January 1, 2014.
Source
scopus
Published In
Proceedings of SPIE
Volume
9033
Publish Date
2014
DOI
10.1117/12.2043829

A knowledge-based approach to improving and homogenizing intensity modulated radiation therapy planning quality among treatment centers: an example application to prostate cancer planning.

PURPOSE: Intensity modulated radiation therapy (IMRT) treatment planning can have wide variation among different treatment centers. We propose a system to leverage the IMRT planning experience of larger institutions to automatically create high-quality plans for outside clinics. We explore feasibility by generating plans for patient datasets from an outside institution by adapting plans from our institution. METHODS AND MATERIALS: A knowledge database was created from 132 IMRT treatment plans for prostate cancer at our institution. The outside institution, a community hospital, provided the datasets for 55 prostate cancer cases, including their original treatment plans. For each "query" case from the outside institution, a similar "match" case was identified in the knowledge database, and the match case's plan parameters were then adapted and optimized to the query case by use of a semiautomated approach that required no expert planning knowledge. The plans generated with this knowledge-based approach were compared with the original treatment plans at several dose cutpoints. RESULTS: Compared with the original plan, the knowledge-based plan had a significantly more homogeneous dose to the planning target volume and a significantly lower maximum dose. The volumes of the rectum, bladder, and femoral heads above all cutpoints were nominally lower for the knowledge-based plan; the reductions were significantly lower for the rectum. In 40% of cases, the knowledge-based plan had overall superior (lower) dose-volume histograms for rectum and bladder; in 54% of cases, the comparison was equivocal; in 6% of cases, the knowledge-based plan was inferior for both bladder and rectum. CONCLUSIONS: Knowledge-based planning was superior or equivalent to the original plan in 95% of cases. The knowledge-based approach shows promise for homogenizing plan quality by transferring planning expertise from more experienced to less experienced institutions.

Authors
Good, D; Lo, J; Lee, WR; Wu, QJ; Yin, F-F; Das, SK
MLA Citation
Good, D, Lo, J, Lee, WR, Wu, QJ, Yin, F-F, and Das, SK. "A knowledge-based approach to improving and homogenizing intensity modulated radiation therapy planning quality among treatment centers: an example application to prostate cancer planning." Int J Radiat Oncol Biol Phys 87.1 (September 1, 2013): 176-181.
PMID
23623460
Source
pubmed
Published In
International Journal of Radiation: Oncology - Biology - Physics
Volume
87
Issue
1
Publish Date
2013
Start Page
176
End Page
181
DOI
10.1016/j.ijrobp.2013.03.015

Development of matched virtual and physical breast phantoms based on patient data

Physical phantoms are essential for the development, optimization, and clinical evaluation of x-ray systems. These phantoms are used for various tests such as quality assurance testing, system characterization, reconstruction evaluation, and dosimetry. They should ideally be capable of serving as ground truth for purposes such as virtual clinical trials. Currently, there is no anthropomorphic 3D physical phantom commercially available. We present our development of a new suite of physical breast phantoms based on real patient data. The phantoms were generated from the NURBS-based extended cardiac-torso (XCAT) breast phantoms, which were segmented from patient dedicated breast computed tomography data. High-resolution multi-material 3D printing technology was used to fabricate the physical models. Glandular tissue and skin were presented by the most radiographically dense photopolymer available to the printer, mimicking a 75% glandular tissue. Adipose tissue was presented by the least radiographically dense photopolymer, mimicking a 35% glandular tissue. The glandular equivalency was measured by comparing x-ray images of samples of the photopolymers available to the printer with those of breast tissue-equivalent materials. The mammographic projections and tomosynthesis reconstructed images of fabricated models showed great improvement over available phantoms, presenting a more realistic breast background. © 2013 SPIE.

Authors
Kiarashi, N; Sturgeon, GM; Nolte, LW; Lo, JY; III, JTD; Segars, WP; Samei, E
MLA Citation
Kiarashi, N, Sturgeon, GM, Nolte, LW, Lo, JY, III, JTD, Segars, WP, and Samei, E. "Development of matched virtual and physical breast phantoms based on patient data." 2013.
Source
scival
Published In
Proceedings of SPIE
Volume
8668
Publish Date
2013
DOI
10.1117/12.2008406

Estimating breast density with dual energy mammography: A simple model based on calibration phantoms

Dual energy digital mammography has been used to suppress specific breast tissue, primarily for the purpose of iodine contrast-enhanced imaging. Another application of dual energy digital mammography is to estimate breast density, as defined by the fraction of glandular tissue, by suppressing adipose tissue. Adipose equivalent phantoms were used to derive the weighting factor for dual energy subtraction at 2, 4, 6, and 8 cm thickness. For each thickness besides 8 cm, measurements were taken over a range of densities (0, 50, and 100%) and used for calibration measurements to model a density map. Once the density map was verified with uniform slabs, the density map was evaluated with 50/50 CIRS 020 phantom at 2, 4, and 6 cm thickness and demonstrated the feasibility of using dual energy subtraction to estimate breast density on complex phantoms. © 2013 SPIE.

Authors
Chung, H; Ikejimba, L; Kiarashi, N; Samei, E; Hoernig, M; Lo, JY
MLA Citation
Chung, H, Ikejimba, L, Kiarashi, N, Samei, E, Hoernig, M, and Lo, JY. "Estimating breast density with dual energy mammography: A simple model based on calibration phantoms." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 8668 (2013).
Source
scival
Published In
Proceedings of SPIE
Volume
8668
Publish Date
2013
DOI
10.1117/12.2008398

Quality assurance and training procedures for computer-aided detection and diagnosis systems in clinical use

Computer-aided detection/diagnosis (CAD) is increasingly used for decision support by clinicians for detection and interpretation of diseases. However, there are no quality assurance (QA) requirements for CAD in clinical use at present. QA of CAD is important so that end users can be made aware of changes in CAD performance both due to intentional or unintentional causes. In addition, end-user training is critical to prevent improper use of CAD, which could potentially result in lower overall clinical performance. Research on QA of CAD and user training are limited to date. The purpose of this paper is to bring attention to these issues, inform the readers of the opinions of the members of the American Association of Physicists in Medicine (AAPM) CAD subcommittee, and thus stimulate further discussion in the CAD community on these topics. The recommendations in this paper are intended to be work items for AAPM task groups that will be formed to address QA and user training issues on CAD in the future. The work items may serve as a framework for the discussion and eventual design of detailed QA and training procedures for physicists and users of CAD. Some of the recommendations are considered by the subcommittee to be reasonably easy and practical and can be implemented immediately by the end users; others are considered to be "best practice" approaches, which may require significant effort, additional tools, and proper training to implement. The eventual standardization of the requirements of QA procedures for CAD will have to be determined through consensus from members of the CAD community, and user training may require support of professional societies. It is expected that high-quality CAD and proper use of CAD could allow these systems to achieve their true potential, thus benefiting both the patients and the clinicians, and may bring about more widespread clinical use of CAD for many other diseases and applications. It is hoped that the awareness of the need for appropriate CAD QA and user training will stimulate new ideas and approaches for implementing such procedures efficiently and effectively as well as funding opportunities to fulfill such critical efforts. © 2013 American Association of Physicists in Medicine.

Authors
Huo, Z; Summers, RM; Paquerault, S; Lo, J; Hoffmeister, J; III, SGA; Freedman, MT; Lin, J; Lo, S-CB; Petrick, N; Sahiner, B; Fryd, D; Yoshida, H; Chan, H-P
MLA Citation
Huo, Z, Summers, RM, Paquerault, S, Lo, J, Hoffmeister, J, III, SGA, Freedman, MT, Lin, J, Lo, S-CB, Petrick, N, Sahiner, B, Fryd, D, Yoshida, H, and Chan, H-P. "Quality assurance and training procedures for computer-aided detection and diagnosis systems in clinical use." Medical Physics 40.7 (2013).
PMID
23822459
Source
scival
Published In
Medical physics
Volume
40
Issue
7
Publish Date
2013
DOI
10.1118/1.4807642

Development of a dynamic 4D anthropomorphic breast phantom for contrast-based breast imaging

Mammography is currently the most widely accepted tool for detection and diagnosis of breast cancer. However, the sensitivity of mammography is reduced in women with dense breast tissue due to tissue overlap, which may obscure lesions. Digital breast tomosynthesis with contrast enhancement reduces tissue overlap and provides additional functional information about lesions (i.e. morphology and kinetics), which in turn may improve lesion characterization. The performance of such techniques is highly dependent on the structural composition of the breast, which varies significantly across patients. Therefore, optimization of breast imaging systems should be done with respect to this patient versatility. Furthermore, imaging techniques that employ contrast require the inclusion of a temporally varying breast composition with respect to the contrast agent kinetics to enable the optimization of the system. To these ends, we have developed a dynamic 4D anthropomorphic breast phantom, which can be used for optimizing a breast imaging system by incorporating material characteristics. The presented dynamic phantom is based on two recently developed anthropomorphic breast phantoms, which can be representative of a whole population through their randomized anatomical feature generation and various compression levels. The 4D dynamic phantom is incorporated with the kinetics of contrast agent uptake in different tissues and can realistically model benign and malignant lesions. To demonstrate the utility of the proposed dynamic phantom, contrast-enhanced digital mammography and breast tomosynthesis were simulated where a ray-tracing algorithm emulated the projections, a filtered back projection algorithm was used for reconstruction, and dual-energy and temporal subtractions were performed and compared. © 2012 Copyright Society of Photo-Optical Instrumentation Engineers (SPIE).

Authors
Kiarashi, N; Lin, Y; Segars, WP; Ghate, SV; Ikejimba, L; Chen, B; Lo, JY; Iii, JTD; Nolte, LW; Samei, E
MLA Citation
Kiarashi, N, Lin, Y, Segars, WP, Ghate, SV, Ikejimba, L, Chen, B, Lo, JY, Iii, JTD, Nolte, LW, and Samei, E. "Development of a dynamic 4D anthropomorphic breast phantom for contrast-based breast imaging." 2012.
Source
scival
Published In
Proceedings of SPIE
Volume
8313
Publish Date
2012
DOI
10.1117/12.913332

Application of a dynamic 4D anthropomorphic breast phantom in contrast-based imaging system optimization: Dual-energy or temporal subtraction?

We previously developed a dynamic 4D anthropomorphic breast phantom, which can be used to optimize contrast-based breast imaging systems, accounting for patient variability and contrast kinetics [1]. In this study we aim to compare the performance of contrast-enhanced mammographic and tomosynthesis imaging protocols followed by temporal subtraction and dual-energy subtraction, qualitatively and quantitatively across a couple of patient models. Signal-difference-to-noise ratio (SDNR) is measured for the six paradigms of contrast enhanced, temporally subtracted, and dual-energy subtracted mammography and tomosynthesis and compared. The results show how the performance is more dependent on the breast model in mammography than in tomosynthesis. Also, it is observed that dual-energy subtraction can be beneficial in mammography, whereas it is not advantageous in tomosynthesis. Lastly, the results suggest that temporal subtraction in general outperforms dual-energy subtraction. © 2012 Springer-Verlag Berlin Heidelberg.

Authors
Kiarashi, N; Ghate, SV; Lo, JY; Nolte, LW; Samei, E
MLA Citation
Kiarashi, N, Ghate, SV, Lo, JY, Nolte, LW, and Samei, E. "Application of a dynamic 4D anthropomorphic breast phantom in contrast-based imaging system optimization: Dual-energy or temporal subtraction?." Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7361 LNCS (2012): 658-665.
Source
scival
Published In
Lecture notes in computer science
Volume
7361 LNCS
Publish Date
2012
Start Page
658
End Page
665
DOI
10.1007/978-3-642-31271-7_85

Task-based strategy for optimized contrast enhanced breast imaging: Analysis of six imaging techniques for mammography and tomosynthesis

Digital breast tomosynthesis (DBT) is a novel x-ray imaging technique that provides 3D structural information of the breast. In contrast to 2D mammography, DBT minimizes tissue overlap potentially improving cancer detection and reducing number of unnecessary recalls. The addition of a contrast agent to DBT and mammography for lesion enhancement has the benefit of providing functional information of a lesion, as lesion contrast uptake and washout patterns may help differentiate between benign and malignant tumors. This study used a task-based method to determine the optimal imaging approach by analyzing six imaging paradigms in terms of their ability to resolve iodine at a given dose: contrast enhanced mammography and tomosynthesis, temporal subtraction mammography and tomosynthesis, and dual energy subtraction mammography and tomosynthesis. Imaging performance was characterized using a detectability index d', derived from the system task transfer function (TTF), an imaging task, iodine contrast, and the noise power spectrum (NPS). The task modeled a 5 mm lesion containing iodine concentrations between 2.1 mg/cc and 8.6 mg/cc. TTF was obtained using an edge phantom, and the NPS was measured over several exposure levels, energies, and target-filter combinations. Using a structured CIRS phantom, d' was generated as a function of dose and iodine concentration. In general, higher dose gave higher d', but for the lowest iodine concentration and lowest dose, dual energy subtraction tomosynthesis and temporal subtraction tomosynthesis demonstrated the highest performance. © 2012 Copyright Society of Photo-Optical Instrumentation Engineers (SPIE).

Authors
Ikejimba, L; Kiarashi, N; Lin, Y; Chen, B; Ghate, SV; Zerhouni, M; Samei, E; Lo, JY
MLA Citation
Ikejimba, L, Kiarashi, N, Lin, Y, Chen, B, Ghate, SV, Zerhouni, M, Samei, E, and Lo, JY. "Task-based strategy for optimized contrast enhanced breast imaging: Analysis of six imaging techniques for mammography and tomosynthesis." 2012.
PMID
24877819
Source
scival
Published In
Proceedings of SPIE
Volume
8313
Publish Date
2012
DOI
10.1117/12.913377

3D biopsy for tomosynthesis: Simulation of prior information based reconstruction for dose and artifact reduction

Accurately targeting of small lesions for success is crucial in breast biopsy. In this paper, we proposed a new 3D tomobased biopsy, which is characterized in being more accurate, easier to perform, lower in dose, and free of metal artifact. In the scout phase, a conventional tomosynthesis scan is performed, and the reconstructed 3D image is then used for radiologists to accurately localize target volume and determine optimized needle path. In the prefire phase, two prefire stereotactic images are obtained at +24° and -24° angular levels for retrieving needle and shifted lesion locations. By combining the reconstructed 3D tomosynthesis image, needle location and lesion location, synthetic prefire and postfire images are generated for radiologists' reference before firing the real needle. The proposed scheme not only improves the biopsy accuracy but also reduces dose by 3.7-5.6 times compared to conventional mammo-based stereotactic biopsy. A simulation using anthropomorphic phantom was conducted to verify our method. Both needle and lesion were precisely recovered just based on two tomo angled images. For the needle registration, the sum of translation discrepancy is less than 3 pixels, and the sum of rotation discrepancy is less than 3 degrees. For the lesion registration, the sum of coordinate discrepancy is less than 4 pixels. The predicted 3D prefire and postfire images exhibited more intuitive spatial relationship of the shifted lesion and biopsy needle tip than mammo-based stereotactic biopsy. © 2012 Copyright Society of Photo-Optical Instrumentation Engineers (SPIE).

Authors
Lin, Y; Ghate, S; Lo, J; Samei, E
MLA Citation
Lin, Y, Ghate, S, Lo, J, and Samei, E. "3D biopsy for tomosynthesis: Simulation of prior information based reconstruction for dose and artifact reduction." 2012.
Source
scival
Published In
Proceedings of SPIE
Volume
8313
Publish Date
2012
DOI
10.1117/12.912346

Mutual information-based template matching scheme for detection of breast masses: from mammography to digital breast tomosynthesis.

Development of a computational decision aid for a new medical imaging modality typically is a long and complicated process. It consists of collecting data in the form of images and annotations, development of image processing and pattern recognition algorithms for analysis of the new images and finally testing of the resulting system. Since new imaging modalities are developed more rapidly than ever before, any effort for decreasing the time and cost of this development process could result in maximizing the benefit of the new imaging modality to patients by making the computer aids quickly available to radiologists that interpret the images. In this paper, we make a step in this direction and investigate the possibility of translating the knowledge about the detection problem from one imaging modality to another. Specifically, we present a computer-aided detection (CAD) system for mammographic masses that uses a mutual information-based template matching scheme with intelligently selected templates. We presented principles of template matching with mutual information for mammography before. In this paper, we present an implementation of those principles in a complete computer-aided detection system. The proposed system, through an automatic optimization process, chooses the most useful templates (mammographic regions of interest) using a large database of previously collected and annotated mammograms. Through this process, the knowledge about the task of detecting masses in mammograms is incorporated in the system. Then, we evaluate whether our system developed for screen-film mammograms can be successfully applied not only to other mammograms but also to digital breast tomosynthesis (DBT) reconstructed slices without adding any DBT cases for training. Our rationale is that since mutual information is known to be a robust inter-modality image similarity measure, it has high potential of transferring knowledge between modalities in the context of the mass detection task. Experimental evaluation of the system on mammograms showed competitive performance compared to other mammography CAD systems recently published in the literature. When the system was applied "as-is" to DBT, its performance was notably worse than that for mammograms. However, with a simple additional preprocessing step, the performance of the system reached levels similar to that obtained for mammograms. In conclusion, the presented CAD system not only performed competitively on screen-film mammograms but it also performed robustly on DBT showing that direct transfer of knowledge across breast imaging modalities for mass detection is in fact possible.

Authors
Mazurowski, MA; Lo, JY; Harrawood, BP; Tourassi, GD
MLA Citation
Mazurowski, MA, Lo, JY, Harrawood, BP, and Tourassi, GD. "Mutual information-based template matching scheme for detection of breast masses: from mammography to digital breast tomosynthesis." J Biomed Inform 44.5 (October 2011): 815-823.
PMID
21554985
Source
pubmed
Published In
Journal of Biomedical Informatics
Volume
44
Issue
5
Publish Date
2011
Start Page
815
End Page
823
DOI
10.1016/j.jbi.2011.04.008

Breast tomosynthesis: state-of-the-art and review of the literature.

Authors
Baker, JA; Lo, JY
MLA Citation
Baker, JA, and Lo, JY. "Breast tomosynthesis: state-of-the-art and review of the literature." Acad Radiol 18.10 (October 2011): 1298-1310. (Review)
PMID
21893296
Source
pubmed
Published In
Academic Radiology
Volume
18
Issue
10
Publish Date
2011
Start Page
1298
End Page
1310
DOI
10.1016/j.acra.2011.06.011

Knowledge-based IMRT treatment planning for prostate cancer.

PURPOSE: To demonstrate the feasibility of using a knowledge base of prior treatment plans to generate new prostate intensity modulated radiation therapy (IMRT) plans. Each new case would be matched against others in the knowledge base. Once the best match is identified, that clinically approved plan is used to generate the new plan. METHODS: A database of 100 prostate IMRT treatment plans was assembled into an information-theoretic system. An algorithm based on mutual information was implemented to identify similar patient cases by matching 2D beam's eye view projections of contours. Ten randomly selected query cases were each matched with the most similar case from the database of prior clinically approved plans. Treatment parameters from the matched case were used to develop new treatment plans. A comparison of the differences in the dose-volume histograms between the new and the original treatment plans were analyzed. RESULTS: On average, the new knowledge-based plan is capable of achieving very comparable planning target volume coverage as the original plan, to within 2% as evaluated for D98, D95, and D1. Similarly, the dose to the rectum and dose to the bladder are also comparable to the original plan. For the rectum, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 are 1.8% +/- 8.5%, -2.5% +/- 13.9%, and -13.9% +/- 23.6%, respectively. For the bladder, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 are -5.9% +/- 10.8%, -12.2% +/- 14.6%, and -24.9% +/- 21.2%, respectively. A negative percentage difference indicates that the new plan has greater dose sparing as compared to the original plan. CONCLUSIONS: The authors demonstrate a knowledge-based approach of using prior clinically approved treatment plans to generate clinically acceptable treatment plans of high quality. This semiautomated approach has the potential to improve the efficiency of the treatment planning process while ensuring that high quality plans are developed.

Authors
Chanyavanich, V; Das, SK; Lee, WR; Lo, JY
MLA Citation
Chanyavanich, V, Das, SK, Lee, WR, and Lo, JY. "Knowledge-based IMRT treatment planning for prostate cancer." Med Phys 38.5 (May 2011): 2515-2522.
Website
http://hdl.handle.net/10161/3879
PMID
21776786
Source
pubmed
Published In
Medical physics
Volume
38
Issue
5
Publish Date
2011
Start Page
2515
End Page
2522
DOI
10.1118/1.3574874

Comparative performance of multiview stereoscopic and mammographic display modalities for breast lesion detection.

PURPOSE: Mammography is known to be one of the most difficult radiographic exams to interpret. Mammography has important limitations, including the superposition of normal tissue that can obscure a mass, chance alignment of normal tissue to mimic a true lesion and the inability to derive volumetric information. It has been shown that stereomammography can overcome these deficiencies by showing that layers of normal tissue lay at different depths. If standard stereomammography (i.e., a single stereoscopic pair consisting of two projection images) can significantly improve lesion detection, how will multiview stereoscopy (MVS), where many projection images are used, compare to mammography? The aim of this study was to assess the relative performance of MVS compared to mammography for breast mass detection. METHODS: The MVS image sets consisted of the 25 raw projection images acquired over an arc of approximately 45 degrees using a Siemens prototype breast tomosynthesis system. The mammograms were acquired using a commercial Siemens FFDM system. The raw data were taken from both of these systems for 27 cases and realistic simulated mass lesions were added to duplicates of the 27 images at the same local contrast. The images with lesions (27 mammography and 27 MVS) and the images without lesions (27 mammography and 27 MVS) were then postprocessed to provide comparable and representative image appearance across the two modalities. All 108 image sets were shown to five full-time breast imaging radiologists in random order on a state-of-the-art stereoscopic display. The observers were asked to give a confidence rating for each image (0 for lesion definitely not present, 100 for lesion definitely present). The ratings were then compiled and processed using ROC and variance analysis. RESULTS: The mean AUC for the five observers was 0.614 +/- 0.055 for mammography and 0.778 +/- 0.052 for multiview stereoscopy. The difference of 0.164 +/- 0.065 was statistically significant with a p-value of 0.0148. CONCLUSIONS: The differences in the AUCs and the p-value suggest that multiview stereoscopy has a statistically significant advantage over mammography in the detection of simulated breast masses. This highlights the dominance of anatomical noise compared to quantum noise for breast mass detection. It also shows that significant lesion detection can be achieved with MVS without any of the artifacts associated with tomosynthesis.

Authors
Webb, LJ; Samei, E; Lo, JY; Baker, JA; Ghate, SV; Kim, C; Soo, MS; Walsh, R
MLA Citation
Webb, LJ, Samei, E, Lo, JY, Baker, JA, Ghate, SV, Kim, C, Soo, MS, and Walsh, R. "Comparative performance of multiview stereoscopic and mammographic display modalities for breast lesion detection." Med Phys 38.4 (April 2011): 1972-1980.
Website
http://hdl.handle.net/10161/2508
PMID
21626930
Source
pubmed
Published In
Medical physics
Volume
38
Issue
4
Publish Date
2011
Start Page
1972
End Page
1980
DOI
10.1118/1.3562901

Computer-aided classification of breast masses: performance and interobserver variability of expert radiologists versus residents.

PURPOSE: To evaluate the interobserver variability in descriptions of breast masses by dedicated breast imagers and radiology residents and determine how any differences in lesion description affect the performance of a computer-aided diagnosis (CAD) computer classification system. MATERIALS AND METHODS: Institutional review board approval was obtained for this HIPAA-compliant study, and the requirement to obtain informed consent was waived. Images of 50 breast lesions were individually interpreted by seven dedicated breast imagers and 10 radiology residents, yielding 850 lesion interpretations. Lesions were described with use of 11 descriptors from the Breast Imaging Reporting and Data System, and interobserver variability was calculated with the Cohen κ statistic. Those 11 features were selected, along with patient age, and merged together by a linear discriminant analysis (LDA) classification model trained by using 1005 previously existing cases. Variability in the recommendations of the computer model for different observers was also calculated with the Cohen κ statistic. RESULTS: A significant difference was observed for six lesion features, and radiology residents had greater interobserver variability in their selection of five of the six features than did dedicated breast imagers. The LDA model accurately classified lesions for both sets of observers (area under the receiver operating characteristic curve = 0.94 for residents and 0.96 for dedicated imagers). Sensitivity was maintained at 100% for residents and improved from 98% to 100% for dedicated breast imagers. For residents, the computer model could potentially improve the specificity from 20% to 40% (P < .01) and the κ value from 0.09 to 0.53 (P < .001). For dedicated breast imagers, the computer model could increase the specificity from 34% to 43% (P = .16) and the κ value from 0.21 to 0.61 (P < .001). CONCLUSION: Among findings showing a significant difference, there was greater interobserver variability in lesion descriptions among residents; however, an LDA model using data from either dedicated breast imagers or residents yielded a consistently high performance in the differentiation of benign from malignant breast lesions, demonstrating potential for improving specificity and decreasing interobserver variability in biopsy recommendations.

Authors
Singh, S; Maxwell, J; Baker, JA; Nicholas, JL; Lo, JY
MLA Citation
Singh, S, Maxwell, J, Baker, JA, Nicholas, JL, and Lo, JY. "Computer-aided classification of breast masses: performance and interobserver variability of expert radiologists versus residents." Radiology 258.1 (January 2011): 73-80.
PMID
20971779
Source
pubmed
Published In
Radiology
Volume
258
Issue
1
Publish Date
2011
Start Page
73
End Page
80
DOI
10.1148/radiol.10081308

Validation of a 3D hidden-Markov model for breast tissue segmentation and density estimation from MR and tomosynthesis images

Breast cancer is the most common non-skin cancer and the second leading cause of cancer-related death in women here in the United States. Mammography is the current standard clinical imaging modality for breast cancer screening and diagnosis, and mammographic breast density (i.e. the percentage of the entire breast volume that is taken up by dense glandular tissue) has been shown to be a biomarker well correlated with cancer risk. However, a mammogram is limited by its projective nature, and its quantitative abilities would likely be surpassed by 3D imaging modalities. This study plans to extract quantitative 3D breast tissue density information using a fully automatic probabilistic model trained on pre-segmented MRIs. This model ground truth was obtained from MRIs for 293 breasts by iterative threshold-based voxel value classification. Before model training/testing, all images were processed to optimize the available range of values for this breast tissue segmentation task. After training a 3D hidden Markov model (HMM) on 10 segmented ground truth MR images, our model was validated by segmenting the remaining 283 breasts. Initial training/testing of the HMM on MRIs matched density to thresholding within 5% for 99/283 breasts and 10% for 171/213 breasts. After the same task-based value optimization method was applied to digital breast tomosynthesis (tomo) images, the same trained HMM was tested to segment tomo volumes of subjects with at least one MRI for validation. HMM segmentation was qualitatively superior at the most cranial/caudal slices in MRIs and quantitatively superior when tested across modalities. Its robustness and ease of modification give the HMM great promise and potential for expansion to more novel 3D breast imaging modalities like breast tomo. © 2011 IEEE.

Authors
Shafer, CM; Seewaldt, VL; Lo, JY
MLA Citation
Shafer, CM, Seewaldt, VL, and Lo, JY. "Validation of a 3D hidden-Markov model for breast tissue segmentation and density estimation from MR and tomosynthesis images." Proceedings of the 2011 Biomedical Sciences and Engineering Conference: Image Informatics and Analytics in Biomedicine, BSEC 2011 (2011).
Source
scival
Published In
Proceedings of the 2011 Biomedical Sciences and Engineering Conference: Image Informatics and Analytics in Biomedicine, BSEC 2011
Publish Date
2011
DOI
10.1109/BSEC.2011.5872317

Segmentation of adipose and glandular tissue for breast tomosynthesis imaging using a 3D hidden-Markov model trained on breast MRIs

Breast tomosynthesis involves a restricted number of images acquired in an arc in conventional mammography projection geometry. Despite its angular undersampling, tomosynthesis projections are reconstructed into a volume at a dose comparable to mammography. Tomosynthesis thus provides depth information, which is especially beneficial to patients with dense breasts. Because the device can be based on an existing FFDM unit, tomosynthesis may be used to accurately assess breast tissue composition, which would greatly benefit high-risk patients with less access to costly imaging modalities such as MRI. This study plans to extract quantitative 3D breast tissue density information using a fully automatic probabilistic model trained on segmented MRIs. The MRI ground truth was obtained for 293 breasts by iterative threshold-based fatty / glandular tissue segmentation. After training a 3D hidden Markov model (HMM) on 10 MR volumes, our model was validated by segmenting 214 of the 293 breasts. After the tomosynthesis value optimization, the same trained HMM was tested to segment breast tomosynthesis volumes of subjects whose MRIs were used for validation. Initial training / testing of the HMM on MRIs matched density to thresholding within 5% for 70/214 breasts and 10% for 127/214 breasts. HMM segmentation was qualitatively superior at the cranial/caudal end slices in MRIs and quantitatively superior for most tested tomosynthesis volumes. Its robustness and ease of modification give the HMM great promise and potential for expansion in this multi-modality study. © 2011 SPIE.

Authors
Shafer, CM; Seewaldt, VL; Lo, JY
MLA Citation
Shafer, CM, Seewaldt, VL, and Lo, JY. "Segmentation of adipose and glandular tissue for breast tomosynthesis imaging using a 3D hidden-Markov model trained on breast MRIs." 2011.
Source
scival
Published In
Proceedings of SPIE
Volume
7961
Publish Date
2011
DOI
10.1117/12.878137

Efficient fourier-wavelet super-resolution.

Super-resolution (SR) is the process of combining multiple aliased low-quality images to produce a high-resolution high-quality image. Aside from registration and fusion of low-resolution images, a key process in SR is the restoration and denoising of the fused images. We present a novel extension of the combined Fourier-wavelet deconvolution and denoising algorithm ForWarD to the multiframe SR application. Our method first uses a fast Fourier-base multiframe image restoration to produce a sharp, yet noisy estimate of the high-resolution image. Our method then applies a space-variant nonlinear wavelet thresholding that addresses the nonstationarity inherent in resolution-enhanced fused images. We describe a computationally efficient method for implementing this space-variant processing that leverages the efficiency of the fast Fourier transform (FFT) to minimize complexity. Finally, we demonstrate the effectiveness of this algorithm for regular imagery as well as in digital mammography.

Authors
Robinson, MD; Toth, CA; Lo, JY; Farsiu, S
MLA Citation
Robinson, MD, Toth, CA, Lo, JY, and Farsiu, S. "Efficient fourier-wavelet super-resolution." IEEE Trans Image Process 19.10 (October 2010): 2669-2681.
PMID
20460208
Source
pubmed
Published In
IEEE Transactions on Image Processing
Volume
19
Issue
10
Publish Date
2010
Start Page
2669
End Page
2681
DOI
10.1109/TIP.2010.2050107

The quantitative potential for breast tomosynthesis imaging.

PURPOSE: Due to its limited angular scan range, breast tomosynthesis has lower resolution in the depth direction, which may limit its accuracy in quantifying tissue density. This study assesses the quantitative potential of breast tomosynthesis using relatively simple reconstruction and image processing algorithms. This quantitation could allow improved characterization of lesions as well as image processing to present tomosynthesis images with the familiar appearance of mammography by preserving more low-frequency information. METHODS: All studies were based on a Siemens prototype MAMMOMAT Novation TOMO breast tomo system with a 45 degrees total angular span. This investigation was performed using both simulations and empirical measurements. Monte Carlo simulations were conducted using the breast tomosynthesis geometry and tissue-equivalent, uniform, voxelized phantoms with cuboid lesions of varying density embedded within. Empirical studies were then performed using tissue-equivalent plastic phantoms which were imaged on the actual prototype system. The material surrounding the lesions was set to either fat-equivalent or glandular-equivalent plastic. From the simulation experiments, the effects of scatter, lesion depth, and background material density were studied. The empirical experiments studied the effects of lesion depth, background material density, x-ray tube energy, and exposure level. Additionally, the proposed analysis methods were independently evaluated using a commercially available QA breast phantom (CIRS Model 11A). All image reconstruction was performed with a filtered backprojection algorithm. Reconstructed voxel values within each slice were corrected to reduce background nonuniformities. RESULTS: The resulting lesion voxel values varied linearly with known glandular fraction (correlation coefficient R2 > 0.90) under all simulated and empirical conditions, including for the independent tests with the QA phantom. Analysis of variance performed on the fit line parameters revealed statistically significant differences between the two different background materials and between 28 kVp and the remaining energies (26, 30, and 32 kVp) for the dense experimental phantom. How ever, no significant differences arose between different energies for the fatty phantom, nor for any of the many other combinations of parameters. CONCLUSIONS: These strong linear relationships suggest that breast tomosynthesis image voxel values, after being corrected by our outlined methods, are highly positively correlated with true tissue density. This consistent linearity implies that breast tomosynthesis imaging indeed has potential to be quantitative.

Authors
Shafer, CM; Samei, E; Lo, JY
MLA Citation
Shafer, CM, Samei, E, and Lo, JY. "The quantitative potential for breast tomosynthesis imaging." Med Phys 37.3 (March 2010): 1004-1016.
PMID
20384236
Source
pubmed
Published In
Medical physics
Volume
37
Issue
3
Publish Date
2010
Start Page
1004
End Page
1016
DOI
10.1118/1.3285038

A technique optimization protocol and the potential for dose reduction in digital mammography.

Digital mammography requires revisiting techniques that have been optimized for prior screen/film mammography systems. The objective of the study was to determine optimized radiographic technique for a digital mammography system and demonstrate the potential for dose reduction in comparison to the clinically established techniques based on screen- film. An objective figure of merit (FOM) was employed to evaluate a direct-conversion amorphous selenium (a-Se) FFDM system (Siemens Mammomat Novation(DR), Siemens AG Medical Solutions, Erlangen, Germany) and was derived from the quotient of the squared signal-difference-to-noise ratio to mean glandular dose, for various combinations of technique factors and breast phantom configurations including kilovoltage settings (23-35 kVp), target/filter combinations (Mo-Mo and W-Rh), breast-equivalent plastic in various thicknesses (2-8 cm) and densities (100% adipose, 50% adipose/50% glandular, and 100% glandular), and simulated mass and calcification lesions. When using a W-Rh spectrum, the optimized FOM results for the simulated mass and calcification lesions showed highly consistent trends with kVp for each combination of breast density and thickness. The optimized kVp ranged from 26 kVp for 2 cm 100% adipose breasts to 30 kVp for 8 cm 100% glandular breasts. The use of the optimized W-Rh technique compared to standard Mo-Mo techniques provided dose savings ranging from 9% for 2 cm thick, 100% adipose breasts, to 63% for 6 cm thick, 100% glandular breasts, and for breasts with a 50% adipose/50% glandular composition, from 12% for 2 cm thick breasts up to 57% for 8 cm thick breasts.

Authors
Ranger, NT; Lo, JY; Samei, E
MLA Citation
Ranger, NT, Lo, JY, and Samei, E. "A technique optimization protocol and the potential for dose reduction in digital mammography." Med Phys 37.3 (March 2010): 962-969.
PMID
20384232
Source
pubmed
Published In
Medical physics
Volume
37
Issue
3
Publish Date
2010
Start Page
962
End Page
969
DOI
10.1118/1.3276732

User modeling for improved computer-aided training in radiology: Initial experience

Although mammography is an efficient screening modality for breast cancer, interpretation of mammographic images is a difficult task and notable variability between radiologists performance has been documented. A significant factor impacting radiologists diagnostic performance is adequate training. In this study we propose a new paradigm for computer-assisted training in radiology based on constructing user models for radiologists-in-training that capture individual error making patterns. Such user models are developed and trained to use image features for prediction of the extent of error made by a particular radiologist for variety of cases and therefore estimate difficulty of different types of cases for that radiologist. The constructed user model can be used to develop a personalized training protocol for the radiologist-in-training that focuses on cases that may pose a particular difficulty to the trainee. We initially demonstrate the concept of building individual user models for the task of breast mass diagnosis. Data collected from three resident observers at Duke University was used for the experiments. The result indicate that the proposed models are capable of learning to distinguish difficult and easy cases for each observer with moderate accuracy which shows promise for the proposed concept. © 2010 Copyright SPIE - The International Society for Optical Engineering.

Authors
Mazurowski, MA; Lo, JY; Tourassi, GD
MLA Citation
Mazurowski, MA, Lo, JY, and Tourassi, GD. "User modeling for improved computer-aided training in radiology: Initial experience." 2010.
Source
scival
Published In
Proceedings of SPIE
Volume
7627
Publish Date
2010
DOI
10.1117/12.843863

Optimized image acquisition for breast tomosynthesis in projection and reconstruction space.

Breast tomosynthesis has been an exciting new development in the field of breast imaging. While the diagnostic improvement via tomosynthesis is notable, the full potential of tomosynthesis has not yet been realized. This may be attributed to the dependency of the diagnostic quality of tomosynthesis on multiple variables, each of which needs to be optimized. Those include dose, number of angular projections, and the total angular span of those projections. In this study, the authors investigated the effects of these acquisition parameters on the overall diagnostic image quality of breast tomosynthesis in both the projection and reconstruction space. Five mastectomy specimens were imaged using a prototype tomosynthesis system. 25 angular projections of each specimen were acquired at 6.2 times typical single-view clinical dose level. Images at lower dose levels were then simulated using a noise modification routine. Each projection image was supplemented with 84 simulated 3 mm 3D lesions embedded at the center of 84 nonoverlapping ROIs. The projection images were then reconstructed using a filtered backprojection algorithm at different combinations of acquisition parameters to investigate which of the many possible combinations maximizes the performance. Performance was evaluated in terms of a Laguerre-Gauss channelized Hotelling observer model-based measure of lesion detectability. The analysis was also performed without reconstruction by combining the model results from projection images using Bayesian decision fusion algorithm. The effect of acquisition parameters on projection images and reconstructed slices were then compared to derive an optimization rule for tomosynthesis. The results indicated that projection images yield comparable but higher performance than reconstructed images. Both modes, however, offered similar trends: Performance improved with an increase in the total acquisition dose level and the angular span. Using a constant dose level and angular span, the performance rolled off beyond a certain number of projections, indicating that simply increasing the number of projections in tomosynthesis may not necessarily improve its performance. The best performance for both projection images and tomosynthesis slices was obtained for 15-17 projections spanning an angular are of approximately 45 degrees--the maximum tested in our study, and for an acquisition dose equal to single-view mammography. The optimization framework developed in this framework is applicable to other reconstruction techniques and other multiprojection systems.

Authors
Chawla, AS; Lo, JY; Baker, JA; Samei, E
MLA Citation
Chawla, AS, Lo, JY, Baker, JA, and Samei, E. "Optimized image acquisition for breast tomosynthesis in projection and reconstruction space." Med Phys 36.11 (November 2009): 4859-4869.
PMID
19994493
Source
pubmed
Published In
Medical physics
Volume
36
Issue
11
Publish Date
2009
Start Page
4859
End Page
4869
DOI
10.1118/1.3231814

Can compression be reduced for breast tomosynthesis? Monte carlo study on mass and microcalcification conspicuity in tomosynthesis.

PURPOSE: To assess, in a voxelized anthropomorphic breast phantom, how the conspicuity of breast masses and microcalcifications may be affected by applying reduced breast compression in tomosynthesis. MATERIALS AND METHODS: A breast tomosynthesis system was modeled by using a Monte Carlo program and a voxelized anthropomorphic breast phantom. The Monte Carlo program created simulated tomosynthesis projection images, which were reconstructed by using filtered back-projection software. Reconstructed images were analyzed for mass and microcalcification conspicuity, or the ratio of the lesion contrast to the anatomic and quantum noise surrounding the lesion. This analysis was performed at two compression levels (standard and 12.5% reduction) and for two breast compression thicknesses (4 and 6 cm). The change in conspicuity was analyzed for significance by using a bootstrap method and a paired Student t test. RESULTS: While keeping the glandular radiation dose constant with respective standard and reduced compression levels, the mean mass conspicuities were 1.39 +/- 0.15 (standard error of the mean) and 1.46 +/- 0.22 for a 4-cm breast compression phantom and 1.26 +/- 0.15 and 1.22 +/- 0.20 for a 6-cm breast phantom, and the mean microcalcification conspicuities were 16.2 +/- 2.87 and 18.6 +/- 2.63 for a 4-cm breast phantom and 11.4 +/- 1.11 and 10.6 +/- 1.18 for a 6-cm breast compression phantom. CONCLUSION: For constant glandular dose, mass and microcalcification conspicuity remained approximately constant with decreased compression. Constant conspicuity implies that reduced compression would have a minimal effect on radiologists' performance, which suggests that there is justification for a measured reduction of breast compression for breast tomosynthesis, increasing the comfort of women undergoing the examination.

Authors
Saunders, RS; Samei, E; Lo, JY; Baker, JA
MLA Citation
Saunders, RS, Samei, E, Lo, JY, and Baker, JA. "Can compression be reduced for breast tomosynthesis? Monte carlo study on mass and microcalcification conspicuity in tomosynthesis." Radiology 251.3 (June 2009): 673-682.
PMID
19474373
Source
pubmed
Published In
Radiology
Volume
251
Issue
3
Publish Date
2009
Start Page
673
End Page
682
DOI
10.1148/radiol.2521081278

Towards optimized acquisition scheme for multiprojection correlation imaging of breast cancer.

RATIONALE AND OBJECTIVES: Correlation imaging (CI) is a form of multiprojection imaging in which multiple images of a patient are acquired from slightly different angles. Information from these images is combined to make the final diagnosis. A critical factor affecting the performance of CI is its data acquisition scheme, because nonoptimized acquisition may distort pathologic indicators. The authors describe a computer-aided detection (CADe) methodology to optimize the acquisition scheme of CI for superior diagnostic accuracy. MATERIALS AND METHODS: Images from 106 subjects were used. For each subject, 25 angular projections of a single breast were acquired. Projection images were supplemented with a simulated 3-mm three-dimensional lesion. Each projection was then processed using a traditional CADe algorithm at high sensitivity, followed by the reduction of false-positives by combining the geometric correlation information available from the multiple images. The performance of the CI system was determined in terms of free-response receiver-operating characteristic curves and the areas under receiver-operating characteristic curves. For optimization, the components of acquisition, such as the number of projections and their angular span, were systematically changed to investigate which of the many possible combinations maximized the obtainable CADe sensitivity and specificity. RESULTS: The performance of the CI system was improved by increasing the angular span. Increasing the number of angular projections beyond a certain number did not improve performance. Maximum performance was obtained between 7 and 10 projections spanning a maximum angular arc of 45 degrees . CONCLUSION: The findings suggest the existence of an optimum acquisition scheme for CI of the breast. CADe results confirmed earlier predictions on the basis of observer models. An optimized CI system may be an important diagnostic tool for improved breast cancer detection.

Authors
Chawla, AS; Saunders, RS; Singh, S; Lo, JY; Samei, E
MLA Citation
Chawla, AS, Saunders, RS, Singh, S, Lo, JY, and Samei, E. "Towards optimized acquisition scheme for multiprojection correlation imaging of breast cancer." Acad Radiol 16.4 (April 2009): 456-463.
PMID
19268858
Source
pubmed
Published In
Academic Radiology
Volume
16
Issue
4
Publish Date
2009
Start Page
456
End Page
463
DOI
10.1016/j.acra.2008.09.013

Do serum biomarkers really measure breast cancer?

BACKGROUND: Because screening mammography for breast cancer is less effective for premenopausal women, we investigated the feasibility of a diagnostic blood test using serum proteins. METHODS: This study used a set of 98 serum proteins and chose diagnostically relevant subsets via various feature-selection techniques. Because of significant noise in the data set, we applied iterated Bayesian model averaging to account for model selection uncertainty and to improve generalization performance. We assessed generalization performance using leave-one-out cross-validation (LOOCV) and receiver operating characteristic (ROC) curve analysis. RESULTS: The classifiers were able to distinguish normal tissue from breast cancer with a classification performance of AUC = 0.82 +/- 0.04 with the proteins MIF, MMP-9, and MPO. The classifiers distinguished normal tissue from benign lesions similarly at AUC = 0.80 +/- 0.05. However, the serum proteins of benign and malignant lesions were indistinguishable (AUC = 0.55 +/- 0.06). The classification tasks of normal vs. cancer and normal vs. benign selected the same top feature: MIF, which suggests that the biomarkers indicated inflammatory response rather than cancer. CONCLUSION: Overall, the selected serum proteins showed moderate ability for detecting lesions. However, they are probably more indicative of secondary effects such as inflammation rather than specific for malignancy.

Authors
Jesneck, JL; Mukherjee, S; Yurkovetsky, Z; Clyde, MA; Marks, JR; Lokshin, AE; Lo, JY
MLA Citation
Jesneck, JL, Mukherjee, S, Yurkovetsky, Z, Clyde, MA, Marks, JR, Lokshin, AE, and Lo, JY. "Do serum biomarkers really measure breast cancer?." BMC cancer 9 (2009): 164-164.
PMID
19476629
Source
manual
Published In
BMC Cancer
Volume
9
Publish Date
2009
Start Page
164
End Page
164
DOI
10.1186/1471-2407-9-164

Computerized 3D breast phantom with enhanced High-Resolution detail

We previously proposed a three-dimensional computerized breast phantom that combines empirical data with the flexibility of mathematical models1. The goal of this project is to enhance the breast phantom to include a more detailed anatomy than currently visible and create additional phantoms from different breast CT data. To improve the level of detail in our existing segmentations, the breast CT data is reconstructed at a higher resolution and additional image processing techniques are used to correct for noise and scatter in the image data. A refined segmentation algorithm is used that incorporates more detail than previously defined. To further enhance high-resolution detail, mathematical models, implementing branching algorithms to extend the glandular tissue throughout the breast and to define Cooper's ligaments, are under investigation. We perform the simulation of mammography and tomosynthesis using an analytical projection algorithm that can be applied directly to the mathematical model of the breast without voxelization 2. This method speeds up image acquisition, reduces voxelization artifacts, and produces higher resolution images than the previously used method. The realistic 3D computerized breast phantom will ultimately be incorporated into the 4DXCAT phantom 3-5 to be used for breast imaging research.©2009 SPIE.

Authors
Li, CM; Segars, WP; Lo, JY; Veress, AI; Boone, JM; III, JTD
MLA Citation
Li, CM, Segars, WP, Lo, JY, Veress, AI, Boone, JM, and III, JTD. "Computerized 3D breast phantom with enhanced High-Resolution detail." 2009.
Source
scival
Published In
Proceedings of SPIE
Volume
7258
Publish Date
2009
DOI
10.1117/12.813529

Optimized lesion detection in breast tomosynthesis

While diagnostic improvement via breast tomosynthesis has been notable, the full potential of tomosynthesis has not yet been realized. This is because of the complex task of optimizing multiple parameters that constitute image acquisition and thus affect tomosynthesis performance. Those parameters include dose, number of angular projections, and the total angular span of those projections. In this study, we investigated the effects of acquisition parameters, independent of each other, on the overall diagnostic image quality of tomosynthesis. Five mastectomy specimens were imaged using a prototype tomosynthesis system. 25 angular projections of each specimen were acquired at 6.2 times typical single-view mammographic dose level. Images at lower dose levels were then simulated using a noise modification routine. Each projection image was supplemented with 84 simulated 3 mm 3D lesions embedded at the center of 84 non-overlapping ROIs. The projection images were then reconstructed using a filtered-back projection (FBP) algorithm at 224 different combinations of acquisition parameters to investigate which one of the many possible combinations maximized performance. Performance was evaluated in terms of a Laguerre-Gauss channelized Hotelling observer model-based measure of lesion detectability. Results showed that performance improved with an increase in the total acquisition dose level and the angular span. At a constant dose level and angular span, the performance rolled-off beyond a certain number of projections, indicating that simply increasing the number of projections in tomosynthesis may not necessarily improve its performance. The best performance was obtained with 15-17 projections spanning an angular arc of ~45° - the maximum tested in our study, and for an acquisition dose equal to single-view mammography. The optimization framework developed in this framework is applicable to other reconstruction techniques and other multi-projection systems. © 2009 SPIE.

Authors
Chawla, AS; Samei, E; Lo, JY
MLA Citation
Chawla, AS, Samei, E, and Lo, JY. "Optimized lesion detection in breast tomosynthesis." 2009.
Source
scival
Published In
Proceedings of SPIE
Volume
7258
Publish Date
2009
DOI
10.1117/12.813964

Automated breast mass detection in 3D reconstructed tomosynthesis volumes: a featureless approach.

The purpose of this study was to propose and implement a computer aided detection (CADe) tool for breast tomosynthesis. This task was accomplished in two stages-a highly sensitive mass detector followed by a false positive (FP) reduction stage. Breast tomosynthesis data from 100 human subject cases were used, of which 25 subjects had one or more mass lesions and the rest were normal. For stage 1, filter parameters were optimized via a grid search. The CADe identified suspicious locations were reconstructed to yield 3D CADe volumes of interest. The first stage yielded a maximum sensitivity of 93% with 7.7 FPs/breast volume. Unlike traditional CADe algorithms in which the second stage FP reduction is done via feature extraction and analysis, instead information theory principles were used with mutual information as a similarity metric. Three schemes were proposed, all using leave-one-case-out cross validation sampling. The three schemes, A, B, and C, differed in the composition of their knowledge base of regions of interest (ROIs). Scheme A's knowledge base was comprised of all the mass and FP ROIs generated by the first stage of the algorithm. Scheme B had a knowledge base that contained information from mass ROIs and randomly extracted normal ROIs. Scheme C had information from three sources of information-masses, FPs, and normal ROIs. Also, performance was assessed as a function of the composition of the knowledge base in terms of the number of FP or normal ROIs needed by the system to reach optimal performance. The results indicated that the knowledge base needed no more than 20 times as many FPs and 30 times as many normal ROIs as masses to attain maximal performance. The best overall system performance was 85% sensitivity with 2.4 FPs per breast volume for scheme A, 3.6 FPs per breast volume for scheme B, and 3 FPs per breast volume for scheme C.

Authors
Singh, S; Tourassi, GD; Baker, JA; Samei, E; Lo, JY
MLA Citation
Singh, S, Tourassi, GD, Baker, JA, Samei, E, and Lo, JY. "Automated breast mass detection in 3D reconstructed tomosynthesis volumes: a featureless approach." Med Phys 35.8 (August 2008): 3626-3636.
PMID
18777923
Source
pubmed
Published In
Medical physics
Volume
35
Issue
8
Publish Date
2008
Start Page
3626
End Page
3636
DOI
10.1118/1.2953562

Optimization of exposure parameters in full field digital mammography.

Optimization of exposure parameters (target, filter, and kVp) in digital mammography necessitates maximization of the image signal-to-noise ratio (SNR), while simultaneously minimizing patient dose. The goal of this study is to compare, for each of the major commercially available full field digital mammography (FFDM) systems, the impact of the selection of technique factors on image SNR and radiation dose for a range of breast thickness and tissue types. This phantom study is an update of a previous investigation and includes measurements on recent versions of two of the FFDM systems discussed in that article, as well as on three FFDM systems not available at that time. The five commercial FFDM systems tested, the Senographe 2000D from GE Healthcare, the Mammomat Novation DR from Siemens, the Selenia from Hologic, the Fischer Senoscan, and Fuji's 5000MA used with a Lorad M-IV mammography unit, are located at five different university test sites. Performance was assessed using all available x-ray target and filter combinations and nine different phantom types (three compressed thicknesses and three tissue composition types). Each phantom type was also imaged using the automatic exposure control (AEC) of each system to identify the exposure parameters used under automated image acquisition. The figure of merit (FOM) used to compare technique factors is the ratio of the square of the image SNR to the mean glandular dose. The results show that, for a given target/filter combination, in general FOM is a slowly changing function of kVp, with stronger dependence on the choice of target/filter combination. In all cases the FOM was a decreasing function of kVp at the top of the available range of kVp settings, indicating that higher tube voltages would produce no further performance improvement. For a given phantom type, the exposure parameter set resulting in the highest FOM value was system specific, depending on both the set of available target/filter combinations, and on the receptor type. In most cases, the AECs of the FFDM systems successfully identified exposure parameters resulting in FOM values near the maximum ones, however, there were several examples where AEC performance could be improved.

Authors
Williams, MB; Raghunathan, P; More, MJ; Seibert, JA; Kwan, A; Lo, JY; Samei, E; Ranger, NT; Fajardo, LL; McGruder, A; McGruder, SM; Maidment, ADA; Yaffe, MJ; Bloomquist, A; Mawdsley, GE
MLA Citation
Williams, MB, Raghunathan, P, More, MJ, Seibert, JA, Kwan, A, Lo, JY, Samei, E, Ranger, NT, Fajardo, LL, McGruder, A, McGruder, SM, Maidment, ADA, Yaffe, MJ, Bloomquist, A, and Mawdsley, GE. "Optimization of exposure parameters in full field digital mammography." Med Phys 35.6 (June 2008): 2414-2423.
PMID
18649474
Source
pubmed
Published In
Medical physics
Volume
35
Issue
6
Publish Date
2008
Start Page
2414
End Page
2423
DOI
10.1118/1.2912177

Neutron-stimulated emission computed tomography of a multi-element phantom.

This paper describes the implementation of neutron-stimulated emission computed tomography (NSECT) for non-invasive imaging and reconstruction of a multi-element phantom. The experimental apparatus and process for acquisition of multi-spectral projection data are described along with the reconstruction algorithm and images of the two elements in the phantom. Independent tomographic reconstruction of each element of the multi-element phantom was performed successfully. This reconstruction result is the first of its kind and provides encouraging proof of concept for proposed subsequent spectroscopic tomography of biological samples using NSECT.

Authors
Floyd, CE; Kapadia, AJ; Bender, JE; Sharma, AC; Xia, JQ; Harrawood, BP; Tourassi, GD; Lo, JY; Crowell, AS; Kiser, MR; Howell, CR
MLA Citation
Floyd, CE, Kapadia, AJ, Bender, JE, Sharma, AC, Xia, JQ, Harrawood, BP, Tourassi, GD, Lo, JY, Crowell, AS, Kiser, MR, and Howell, CR. "Neutron-stimulated emission computed tomography of a multi-element phantom." Phys Med Biol 53.9 (May 7, 2008): 2313-2326.
PMID
18421119
Source
pubmed
Published In
Physics in Medicine and Biology
Volume
53
Issue
9
Publish Date
2008
Start Page
2313
End Page
2326
DOI
10.1088/0031-9155/53/9/008

Dedicated breast computed tomography: volume image denoising via a partial-diffusion equation based technique.

Dedicated breast computed tomography (CT) imaging possesses the potential for improved lesion detection over conventional mammograms, especially for women with dense breasts. The breast CT images are acquired with a glandular dose comparable to that of standard two-view mammography for a single breast. Due to dose constraints, the reconstructed volume has a non-negligible quantum noise when thin section CT slices are visualized. It is thus desirable to reduce noise in the reconstructed breast volume without loss of spatial resolution. In this study, partial diffusion equation (PDE) based denoising techniques specifically for breast CT were applied at different steps along the reconstruction process and it was found that denoising performed better when applied to the projection data rather than reconstructed data. Simulation results from the contrast detail phantom show that the PDE technique outperforms Wiener denoising as well as adaptive trimmed mean filter. The PDE technique increases its performance advantage relative to Wiener techniques when the photon fluence is reduced. With the PDE technique, the sensitivity for lesion detection using the contrast detail phantom drops by less than 7% when the dose is cut down to 40% of the two-view mammography. For subjective evaluation, the PDE technique was applied to two human subject breast data sets acquired on a prototype breast CT system. The denoised images had appealing visual characteristics with much lower noise levels and improved tissue textures while maintaining sharpness of the original reconstructed volume.

Authors
Xia, JQ; Lo, JY; Yang, K; Floyd, CE; Boone, JM
MLA Citation
Xia, JQ, Lo, JY, Yang, K, Floyd, CE, and Boone, JM. "Dedicated breast computed tomography: volume image denoising via a partial-diffusion equation based technique." Med Phys 35.5 (May 2008): 1950-1958.
PMID
18561671
Source
pubmed
Published In
Medical physics
Volume
35
Issue
5
Publish Date
2008
Start Page
1950
End Page
1958
DOI
10.1118/1.2903436

A mathematical model platform for optimizing a multiprojection breast imaging system.

Multiprojection imaging is a technique in which a plurality of digital radiographic images of the same patient are acquired within a short interval of time from slightly different angles. Information from each image is combined to determine the final diagnosis. Projection data are either reconstructed into slices as in the case of tomosynthesis or analyzed directly as in the case of multiprojection correlation imaging technique, thereby avoiding reconstruction artifacts. In this study, the authors investigated the optimum geometry of acquisitions of a multiprojection breast correlation imaging system in terms of the number of projections and their total angular span that yield maximum performance in a task that models clinical decision. Twenty-five angular projections of each breast from 82 human subjects in our breast tomosynthesis database were each supplemented with a simulated 3 mm mass. An approach based on Laguerre-Gauss channelized Hotelling observer was developed to assess the detectability of the mass in terms of receiver operating characteristic (ROC) curves. Two methodologies were developed to integrate results from individual projections into one combined ROC curve as the overall figure of merit. To optimize the acquisition geometry, different components of acquisitions were changed to investigate which one of the many possible configurations maximized the area under the combined ROC curve. Optimization was investigated under two acquisition dose conditions corresponding to a fixed total dose delivered to the patient and a variable dose condition, based on the number of projections used. In either case, the detectability was dependent on the number of projections used, the total angular span of those projections, and the acquisition dose level. In the first case, the detectability approximately followed a bell curve as a function of the number of projections with the maximum between 8 and 16 projections spanning angular arcs of about 23 degrees-45 degrees, respectively. In the second case, the detectability increased with the number of projections approaching an asymptote at 11-17 projections for an angular span of about 45 degrees. These results indicate the inherent information content of the multi-projection image data reflecting the relative role of quantum and anatomical noise in multiprojection breast imaging. The optimization scheme presented here may be applied to any multiprojection imaging modalities and may be extended by including reconstruction in the case of digital breast tomosynthesis and breast computed tomography.

Authors
Chawla, AS; Samei, E; Saunders, RS; Lo, JY; Baker, JA
MLA Citation
Chawla, AS, Samei, E, Saunders, RS, Lo, JY, and Baker, JA. "A mathematical model platform for optimizing a multiprojection breast imaging system." Med Phys 35.4 (April 2008): 1337-1345.
PMID
18491528
Source
pubmed
Published In
Medical physics
Volume
35
Issue
4
Publish Date
2008
Start Page
1337
End Page
1345
DOI
10.1118/1.2885367

Training neural network classifiers for medical decision making: the effects of imbalanced datasets on classification performance.

This study investigates the effect of class imbalance in training data when developing neural network classifiers for computer-aided medical diagnosis. The investigation is performed in the presence of other characteristics that are typical among medical data, namely small training sample size, large number of features, and correlations between features. Two methods of neural network training are explored: classical backpropagation (BP) and particle swarm optimization (PSO) with clinically relevant training criteria. An experimental study is performed using simulated data and the conclusions are further validated on real clinical data for breast cancer diagnosis. The results show that classifier performance deteriorates with even modest class imbalance in the training data. Further, it is shown that BP is generally preferable over PSO for imbalanced training data especially with small data sample and large number of features. Finally, it is shown that there is no clear preference between oversampling and no compensation approach and some guidance is provided regarding a proper selection.

Authors
Mazurowski, MA; Habas, PA; Zurada, JM; Lo, JY; Baker, JA; Tourassi, GD
MLA Citation
Mazurowski, MA, Habas, PA, Zurada, JM, Lo, JY, Baker, JA, and Tourassi, GD. "Training neural network classifiers for medical decision making: the effects of imbalanced datasets on classification performance." Neural Netw 21.2-3 (March 2008): 427-436.
PMID
18272329
Source
pubmed
Published In
Neural Networks
Volume
21
Issue
2-3
Publish Date
2008
Start Page
427
End Page
436
DOI
10.1016/j.neunet.2007.12.031

Point/Counterpoint. Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer.

Authors
Karellas, A; Lo, JY; Orton, CG
MLA Citation
Karellas, A, Lo, JY, and Orton, CG. "Point/Counterpoint. Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer." Med Phys 35.2 (February 2008): 409-411.
PMID
18383660
Source
pubmed
Published In
Medical physics
Volume
35
Issue
2
Publish Date
2008
Start Page
409
End Page
411
DOI
10.1118/1.2825612

Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer

Authors
Karellas, A; Lo, JY; Orton, CG
MLA Citation
Karellas, A, Lo, JY, and Orton, CG. "Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer." MEDICAL PHYSICS 35.2 (February 2008): 409-411.
Source
wos-lite
Published In
Medical physics
Volume
35
Issue
2
Publish Date
2008
Start Page
409
End Page
411
DOI
10.1118/l.2825612

Three-dimensional computer generated breast phantom based on empirical data

The goal of this work is to create a detailed three-dimensional (3D) digital breast phantom based on empirical data and to incorporate it into the four-dimensional (4D) NCAT phantom, a computerized model of the human anatomy widely used in imaging research. Twenty sets of high-resolution breast CT data were used to create anatomically diverse models. The datasets were segmented using techniques developed in our laboratory and the breast structures will be defined using a combination of non-uniform rational b-splines (NURBS) and subdivision surfaces (SD). Imaging data from various modalities (x-ray and nuclear medicine) were simulated to demonstrate the utility of the new breast phantoms. As a proof of concept, a simple compression technique was used to deform the breast models while maintaining a constant volume to simulate modalities (mammography and tomosynthesis) that involve compression. Initial studies using one CT dataset indicate that the simulated breast phantom is capable of providing a realistic and flexible representation of breast tissue and can be used with different acquisition methods to test varying imaging parameters such as dose, resolution, and patient motion. The final model will have a more accurate depiction of the internal breast structures and will be scaleable in terms of size and density. Also, more realistic finite-element techniques will be used to simulate compression. With the ability to simulate realistic, predictive patient imaging data, we believe the phantom will provide a vital tool to investigate current and emerging breast imaging methods and techniques.

Authors
Li, CM; Segars, WP; Lo, JY; Veress, AI; Boone, JM; III, JTD
MLA Citation
Li, CM, Segars, WP, Lo, JY, Veress, AI, Boone, JM, and III, JTD. "Three-dimensional computer generated breast phantom based on empirical data." 2008.
Website
http://hdl.handle.net/10161/3119
Source
scival
Published In
Proceedings of SPIE
Volume
6913
Publish Date
2008
DOI
10.1117/12.772185

Efficient restoration and enhancement of super-resolved X-ray images

Our previous work demonstrates the ability to reconstruct a single higher resolution image from fusing a collection of multiple extremely low-dosage aliased X-ray images. While this computationally efficient method eliminates aliasing artifacts associated with undersampling, it does not address the problem of deblurring the reconstructed image. In this paper, we present a fast nonlinear deblurring algorithm, specifically designed to address the nonstationary noise associated with multiframe reconstructed images. The algorithm uses a combination of Fourier sharpening and wavelet denoising similar to the ForWarD algorithm. Experimental results on enhancing digital mammogram images attest to the effectiveness of the presented method. © 2008 IEEE.

Authors
Robinson, MD; Farsiu, S; Lo, JY; Toth, CA
MLA Citation
Robinson, MD, Farsiu, S, Lo, JY, and Toth, CA. "Efficient restoration and enhancement of super-resolved X-ray images." 2008. 629-632.
Source
scival
Publish Date
2008
Start Page
629
End Page
632
DOI
10.1109/ICIP.2008.4711833

Mass detectability in dedicated breast CT: A simulation study with the application of volume noise removal

Dedicated breast Computed Tomography (CT) is an emerging new technique for breast cancer imaging. Breast CT data can be acquired at a dose level as low as the conventional two-view mammography. Since the dose is equally split into hundreds of projection views, each projection image contains non-ignorable quantum noise. This study is aimed at investigating how volume noise removal affects the mass detectability in breast CT. A Partial Diffusion Equation (PDE) based denoising technique was applied before the reconstruction of either a simulated breast volume embedded with a contrast- detail mass phantom or a real human subject breast CT volume embedded with a simulated spherical mass. By applying a mathematical observer, it is found that the PDE volume noise removal technique improves the mass detectability in breast CT in a statistically significant sense.

Authors
Xia, JQ; Lo, JY
MLA Citation
Xia, JQ, and Lo, JY. "Mass detectability in dedicated breast CT: A simulation study with the application of volume noise removal." 2008.
Source
scival
Published In
8th IEEE International Conference on BioInformatics and BioEngineering, BIBE 2008
Publish Date
2008
DOI
10.1109/BIBE.2008.4696788

Multi-projection correlation imaging as a new diagnostic tool for improved breast cancer detection

Multi-projection imaging technique offers an advantage over single projection imaging techniques in rendering pathology that may be surrounded by a complex cloud of anatomical structures. The process of harnessing the geometrical and statistical dependences between the multiple images available in a multi-projection system to determine the final diagnosis is termed Correlation Imaging (CI). In this study, we are investigating the potential improvement in breast cancer detection via CI. As a key step towards that, the acquisition scheme of CI was first optimized to maximize its diagnostic performance. Toward that end, first a clinically-realistic task was designed and each component of acquisition, namely, the acquisition dose level, the number of projections, and their angular span was systematically changed to determine a specific combination that yielded maximum performance in that task. Finally, the performance of the optimized system was compared with that of standard planar mammography. The results indicated that the performance of CI may potentially be optimized between 15-17 projections spanning an angular arc of 45 o . This optimum performance further improved with increasing dose levels; however, at dose level comparable to mammography, CI provided a factor of 1.1 improvement over mammography. The framework developed in this study to evaluate multi-projections system may be applied to any other multi-projection imaging modality, and by including reconstruction, may be extended to digital breast tomosynthesis and breast computed tomography. © 2008 Springer-Verlag Berlin Heidelberg.

Authors
Chawla, AS; Samei, E; Lo, JY; Mertelmeier, T
MLA Citation
Chawla, AS, Samei, E, Lo, JY, and Mertelmeier, T. "Multi-projection correlation imaging as a new diagnostic tool for improved breast cancer detection." 2008.
Source
scival
Published In
Lecture notes in computer science
Volume
5116 LNCS
Publish Date
2008
Start Page
635
End Page
642
DOI
10.1007/978-3-540-70538-3_88

Knowledge transfer across breast cancer screening modalities: A pilot study using an information-theoretic CADe system for mass detection

We have performed a series of experiments to assess whether a featureless, knowledge-based CADe system that relies on information theoretic principles is capable of transferring knowledge across cases acquired with different imaging modalities. Typical feature-based CADe systems are developed and carefully optimized for a specific imaging modality and platform, namely for screen-film mammograms (SFMs) digitized with a specific digitizer, or for full-field digital mammograms (FFDMs), or for the newly introduced digital breast tomosynthesis (DBT) modality. Multiplatform application of such CADe systems is often limited due to image processing steps that are tailored to the imaging modality and acquisition protocol. It is desirable however to develop CADe systems with the ability to adapt to a dynamically changing environment (i.e., imaging modality) and provide an accurate decision while capitalizing on knowledge acquired at different, yet related environments. Working towards this goal, we present a pilot study using a knowledge-based CADe system for mass detection that uses information theory to assess the similarity between a query and a knowledge case. We evaluate the system's ability to transfer knowledge across three imaging modalities (SFMs digitized with two different digitizers, FFDMs, and DBTs) when performing the detection task. Overall our study showed that although blind translation of the system in a new modality for which no prior knowledge exists results in reduced performance, the system is still able to operate at a very decent level. When the system operated with a knowledge database of mixed cases, its performance was robust and comparable to what observed independently. © 2008 Springer-Verlag Berlin Heidelberg.

Authors
Tourassi, GD; Sharma, AC; Singh, S; Saunders, RS; Lo, JY; Samei, E; Harrawood, BP
MLA Citation
Tourassi, GD, Sharma, AC, Singh, S, Saunders, RS, Lo, JY, Samei, E, and Harrawood, BP. "Knowledge transfer across breast cancer screening modalities: A pilot study using an information-theoretic CADe system for mass detection." 2008.
Source
scival
Published In
Lecture notes in computer science
Volume
5116 LNCS
Publish Date
2008
Start Page
292
End Page
298
DOI
10.1007/978-3-540-70538-3_41

Assessment of low energies and slice depth in the quantification of breast tomosynthesis

This study attempts to assess the quantitative potential of breast tomosynthesis imaging. Tomosynthesis might be a feasible replacement for digital mammography, so it is worthwhile to consider whether it can be quantitative like computed tomography (CT), where the image pixel values are expressed in Hounsfield units. For this investigation, plastic tissue-equivalent breast phantoms with 10 lesions of varying density in the center along with a small density calibration phantom of 5 density-varying lesions were imaged under several different conditions. The measured voxel value for each lesion from a reconstructed slice was linearly rescaled based on the calibration phantom and then plotted against the known glandular fraction of each lesion. It was found that the two different energies and the three different lesion depths all produced linear voxel values versus glandularity relationships. Therefore, tomosynthesis has quantitative potential. However, in order to convert each 3D image's voxel values to values that can be interpreted as a certain glandular fraction, one must consider the x-ray tube energy, slice depth, and many other factors of the imaging system and the breast. © 2008 Springer-Verlag Berlin Heidelberg.

Authors
Shafer, CM; Samei, E; Mertelmeier, T; Saunders, RS; Zerhouni, M; Lo, JY
MLA Citation
Shafer, CM, Samei, E, Mertelmeier, T, Saunders, RS, Zerhouni, M, and Lo, JY. "Assessment of low energies and slice depth in the quantification of breast tomosynthesis." 2008.
Source
scival
Published In
Lecture notes in computer science
Volume
5116 LNCS
Publish Date
2008
Start Page
530
End Page
536
DOI
10.1007/978-3-540-70538-3_74

Effect of similarity metrics and ROI sizes in featureless computer aided detection of breast masses in tomosynthesis

Tomosynthesis as a technique is being developed and studied with the goal of overcoming mammography's limitations due to overlying tissue. Various algorithms exist for tomosynthesis datasets including a novel Computer Aided Detection (CADe) algorithm using a featureless False Positive (FP) reduction stage. The goal of this project is to study the previously unexplored effects of variation of Region of Interest (ROI) sizes as well as the crucial similarity metrics for such a CADe algorithm's performance. Four datasets consisting of 1479 tomosynthesis ROIs were generated by a CADe algorithm from reconstructed volumes of one hundred subjects consisting of 4 different sizes - 128 x 128, 256 x 256, 512 x 512, and 1024 x 1024 pixels. Five different similarity metrics - (1) mutual information, (2) average conditional entropy, (3) joint entropy, (3) Jensen divergence and (4) average Kullback-Leibler divergence were used for the task of FP reduction using a leave-one-case-out sampling scheme. Mutual information and average conditional entropy were the best performing metrics with an Area Under Curve (AUC) of 0.88. Cross-bin measures performed consistently higher than those that rely on only marginal distributions. Also, for all metrics, the datatset consisting of 256 x 256 pixel ROIs gave the best performance. In conclusion, for the tomosynthesis dataset, cross-bin measures such as MI and average conditional entropy should be used over other metrics using a ROI size of 256 x 256 pixels. © 2008 Springer-Verlag Berlin Heidelberg.

Authors
Singh, S; Tourassi, GD; Lo, JY
MLA Citation
Singh, S, Tourassi, GD, and Lo, JY. "Effect of similarity metrics and ROI sizes in featureless computer aided detection of breast masses in tomosynthesis." 2008.
Source
scival
Published In
Lecture notes in computer science
Volume
5116 LNCS
Publish Date
2008
Start Page
286
End Page
291
DOI
10.1007/978-3-540-70538-3_40

Optimized acquisition scheme for multi-projection correlation imaging of breast cancer

We are reporting the optimized acquisition scheme of multi-projection breast Correlation Imaging (CI) technique, which was pioneered in our lab at Duke University. CI is similar to tomosynthesis in its image acquisition scheme. However, instead of analyzing the reconstructed images, the projection images are directly analyzed for pathology. Earlier, we presented an optimized data acquisition scheme for CI using mathematical observer model. In this article, we are presenting a Computer Aided Detection (CADe)-based optimization methodology. Towards that end, images from 106 subjects recruited for an ongoing clinical trial for tomosynthesis were employed. For each patient, 25 angular projections of each breast were acquired. Projection images were supplemented with a simulated 3 mm 3D lesion. Each projection was first processed by a traditional CADe algorithm at high sensitivity, followed by a reduction of false positives by combining geometrical correlation information available from the multiple images. Performance of the CI system was determined in terms of free-response receiver operating characteristics (FROC) curves and the area under ROC curves. For optimization, the components of acquisition such as the number of projections, and their angular span were systematically changed to investigate which one of the many possible combinations maximized the sensitivity and specificity. Results indicated that the performance of the CI system may be maximized with 7-11 projections spanning an angular arc of 44.8°, confirming our earlier findings using observer models. These results indicate that an optimized CI system may potentially be an important diagnostic tool for improved breast cancer detection.

Authors
Chawla, AS; Samei, E; Saunders, RS; Lo, JY; Singh, S
MLA Citation
Chawla, AS, Samei, E, Saunders, RS, Lo, JY, and Singh, S. "Optimized acquisition scheme for multi-projection correlation imaging of breast cancer." 2008.
Source
scival
Published In
Proceedings of SPIE
Volume
6915
Publish Date
2008
DOI
10.1117/12.773174

Computer aided detection of breast masses in tomosynthesis reconstructed volumes using information-theoretic similarity measures

The purpose of this project is to study two Computer Aided Detection (CADe) systems for breast masses for digital tomosynthesis using reconstructed slices. This study used eighty human subject cases collected as part of on-going clinical trials at Duke University. Raw projections images were used to identify suspicious regions in the algorithm's high sensitivity, low specificity stage using a Difference of Gaussian filter. The filtered images were thresholded to yield initial CADe hits that were then shifted and added to yield a 3D distribution of suspicious regions. The initial system performance was 95% sensitivity at 10 false positives per breast volume. Two CADe systems were developed. In system A, the central slice located at the centroid depth was used to extract a 256× 256 Regions of Interest (ROI) database centered at the lesion coordinates. For system B, 5 slices centered at the lesion coordinates were summed before the extraction of 256× 256 ROIs. To avoid issues associated with feature extraction, selection, and merging, information theory principles were used to reduce false positives for both the systems resulting in a classifier performance of 0.81 and 0.865 Area Under Curve (AUC) with leave-one-case-out sampling. This resulted in an overall system performance of 87% sensitivity with 6.1 FPs/volume and 85% sensitivity with 3.8 FPs/ volume for systems A and B respectively. This system therefore has the potential to detect breast masses in tomosynthesis data sets.

Authors
Singh, S; Tourassi, GD; Chawla, AS; Saunders, RS; Samei, E; Lo, JY
MLA Citation
Singh, S, Tourassi, GD, Chawla, AS, Saunders, RS, Samei, E, and Lo, JY. "Computer aided detection of breast masses in tomosynthesis reconstructed volumes using information-theoretic similarity measures." 2008.
Source
scival
Published In
Proceedings of SPIE
Volume
6915
Publish Date
2008
DOI
10.1117/12.772978

Toward quantification of breast tomosynthesis imaging

Due to the high prevalence of breast cancer among women, much is being done to detect breast cancer earlier and more accurately. In current clinical practice, the most widely-used mode of breast imaging is mammography. Its main advantages are high sensitivity and low patient dose, although it is still merely a two-dimensional projection of a three-dimensional object. In digital breast tomosynthesis, a three-dimensional image of the breast can be reconstructed, but x-ray projection images of the breast are taken over a limited angular span. However, the breast tomosynthesis device itself is more similar to a digital mammography system and thus is a feasible replacement for mammography. Because of the angular undersampling in breast tomosynthesis, the reconstructed images are not considered quantitative, so a worthwhile question to answer would be whether the voxel values (VVs) in breast tomosynthesis images can be made to indicate tissue type as Hounsfield units do in CT. through some image processing scheme. To investigate this, simple phantoms were imaged consisting of layers of uniform, tissue-equivalent plastic for the background sandwiching a layer of interest containing multiple, small cuboids of tissue-equivalent plastic. After analyzing the reconstructed tomosynthesis images, it was found that the VV in each lesion increases linearly with tissue glandularity. However, for the two different x-ray tube energies and for the two different beam exposure levels tested, the trend-lines all have different slopes and y-intercepts. Thus, breast tomosynthesis has a definite potential to be quantitative, and it would be worthwhile to study other possible dependent parameters (phantom thickness, overall density, etc.) as well as alternative reconstruction algorithms.

Authors
Shafer, CM; Samei, E; Saunders, RS; Zerhouni, M; Lo, JY
MLA Citation
Shafer, CM, Samei, E, Saunders, RS, Zerhouni, M, and Lo, JY. "Toward quantification of breast tomosynthesis imaging." 2008.
Source
scival
Published In
Proceedings of SPIE
Volume
6913
Publish Date
2008
DOI
10.1117/12.772753

Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer

Authors
Karellas, A; Lo, JY; Orton, CG
MLA Citation
Karellas, A, Lo, JY, and Orton, CG. "Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in imaging the breast and delineating cancer." Medical Physics 35.2 (2008): 409-411.
Source
scival
Published In
Medical physics
Volume
35
Issue
2
Publish Date
2008
Start Page
409
End Page
411
DOI
10.1118/1.2825612

Efficient Restoration and Enhancement of Super-resolved X-ray Images

Authors
Robinson, MD; Farsiu, S; Lo, JY; Toth, CA; IEEE,
MLA Citation
Robinson, MD, Farsiu, S, Lo, JY, Toth, CA, and IEEE, . "Efficient Restoration and Enhancement of Super-resolved X-ray Images." 2008.
Source
wos-lite
Published In
Proceedings / ICIP ... International Conference on Image Processing
Publish Date
2008
Start Page
633
End Page
+

Impulse response and Modulation Transfer Function analysis for Shift-And-Add and Back Projection image reconstruction algorithms in Digital Breast Tomosynthesis (DBT).

Breast cancer is second only to lung cancer as the leading cause of non-preventable cancer death in women. Digital Breast Tomosynthesis (DBT) is a promising technique to improve early breast cancer detection. In this paper, we present the impulse response and Modulation Transfer Function (MTF) analysis to quantitatively compare Shift-And-Add (SAA) and point-by-point Back Projection (BP) three-dimensional image reconstruction algorithms in DBT. A Filtered Back Projection (FBP) deblurring algorithm based on point-by-point BP was used to demonstrate deblurred tomosynthesis images.

Authors
Chen, Y; Lo, JY; Dobbins, JT
MLA Citation
Chen, Y, Lo, JY, and Dobbins, JT. "Impulse response and Modulation Transfer Function analysis for Shift-And-Add and Back Projection image reconstruction algorithms in Digital Breast Tomosynthesis (DBT)." Int J Funct Inform Personal Med 1.2 (2008): 189-204.
PMID
23935707
Source
pubmed
Published In
International Journal of Functional Informatics and Personalised Medicine
Volume
1
Issue
2
Publish Date
2008
Start Page
189
End Page
204
DOI
10.1504/IJFIPM.2008.020187

Importance of point-by-point back projection correction for isocentric motion in digital breast tomosynthesis: relevance to morphology of structures such as microcalcifications.

Digital breast tomosynthesis is a three-dimensional imaging technique that provides an arbitrary set of reconstruction planes in the breast from a limited-angle series of projection images acquired while the x-ray tube moves. Traditional shift-and-add (SAA) tomosynthesis reconstruction is a common mathematical method to line up each projection image based on its shifting amount to generate reconstruction slices. With parallel-path geometry of tube motion, the path of the tube lies in a plane parallel to the plane of the detector. The traditional SAA algorithm gives shift amounts for each projection image calculated only along the direction of x-ray tube movement. However, with the partial isocentric motion of the x-ray tube in breast tomosynthesis, small objects such as microcalcifications appear blurred (for instance, about 1-4 pixels in blur for a microcalcification in a human breast) in traditional SAA images in the direction perpendicular to the direction of tube motion. Some digital breast tomosynthesis algorithms reported in the literature utilize a traditional one-dimensional SAA method that is not wholly suitable for isocentric motion. In this paper, a point-by-point back projection (BP) method is described and compared with traditional SAA for the important clinical task of evaluating morphology of small objects such as microcalcifications. Impulse responses at different three-dimensional locations with five different combinations of imaging acquisition parameters were investigated. Reconstruction images of microcalcifications in a human subject were also evaluated. Results showed that with traditional SAA and 45 degrees view angle of tube movement with respect to the detector, at the same height above the detector, the in-plane blur artifacts were obvious for objects farther away from x-ray source. In a human subject, the appearance of calcifications was blurred in the direction orthogonal to the tube motion with traditional SAA. With point-by-point BP, the appearance of calcifications was sharper. The point-by-point BP method demonstrated improved rendition of microcalcifications in the direction perpendicular to the tube motion direction. With wide angles or for imaging of larger breasts, this point-by-point BP rather than the traditional SAA should also be considered as the basis of further deblurring algorithms that work in conjunction with the BP method.

Authors
Chen, Y; Lo, JY; Dobbins, JT
MLA Citation
Chen, Y, Lo, JY, and Dobbins, JT. "Importance of point-by-point back projection correction for isocentric motion in digital breast tomosynthesis: relevance to morphology of structures such as microcalcifications." Med Phys 34.10 (October 2007): 3885-3892.
PMID
17985634
Source
pubmed
Published In
Medical physics
Volume
34
Issue
10
Publish Date
2007
Start Page
3885
End Page
3892
DOI
10.1118/1.2776256

Information-theoretic CAD system in mammography: entropy-based indexing for computational efficiency and robust performance.

We have previously presented a knowledge-based computer-assisted detection (KB-CADe) system for the detection of mammographic masses. The system is designed to compare a query mammographic region with mammographic templates of known ground truth. The templates are stored in an adaptive knowledge database. Image similarity is assessed with information theoretic measures (e.g., mutual information) derived directly from the image histograms. A previous study suggested that the diagnostic performance of the system steadily improves as the knowledge database is initially enriched with more templates. However, as the database increases in size, an exhaustive comparison of the query case with each stored template becomes computationally burdensome. Furthermore, blind storing of new templates may result in redundancies that do not necessarily improve diagnostic performance. To address these concerns we investigated an entropy-based indexing scheme for improving the speed of analysis and for satisfying database storage restrictions without compromising the overall diagnostic performance of our KB-CADe system. The indexing scheme was evaluated on two different datasets as (i) a search mechanism to sort through the knowledge database, and (ii) a selection mechanism to build a smaller, concise knowledge database that is easier to maintain but still effective. There were two important findings in the study. First, entropy-based indexing is an effective strategy to identify fast a subset of templates that are most relevant to a given query. Only this subset could be analyzed in more detail using mutual information for optimized decision making regarding the query. Second, a selective entropy-based deposit strategy may be preferable where only high entropy cases are maintained in the knowledge database. Overall, the proposed entropy-based indexing scheme was shown to reduce the computational cost of our KB-CADe system by 55% to 80% while maintaining the system's diagnostic performance.

Authors
Tourassi, GD; Harrawood, B; Singh, S; Lo, JY
MLA Citation
Tourassi, GD, Harrawood, B, Singh, S, and Lo, JY. "Information-theoretic CAD system in mammography: entropy-based indexing for computational efficiency and robust performance." Med Phys 34.8 (August 2007): 3193-3204.
PMID
17879782
Source
pubmed
Published In
Medical physics
Volume
34
Issue
8
Publish Date
2007
Start Page
3193
End Page
3204
DOI
10.1118/1.2751075

Breast mass lesions: computer-aided diagnosis models with mammographic and sonographic descriptors.

PURPOSE: To retrospectively develop and evaluate computer-aided diagnosis (CAD) models that include both mammographic and sonographic descriptors. MATERIALS AND METHODS: Institutional review board approval was obtained for this HIPAA-compliant study. A waiver of informed consent was obtained. Mammographic and sonographic examinations were performed in 737 patients (age range, 17-87 years), which yielded 803 breast mass lesions (296 malignant, 507 benign). Radiologist-interpreted features from mammograms and sonograms were used as input features for linear discriminant analysis (LDA) and artificial neural network (ANN) models to differentiate benign from malignant lesions. An LDA with all the features was compared with an LDA with only stepwise-selected features. Classification performances were quantified by using receiver operating characteristic (ROC) analysis and were evaluated in a train, validate, and retest scheme. On the retest set, both LDAs were compared with radiologist assessment score of malignancy. RESULTS: Both the LDA and ANN achieved high classification performance with cross validation (area under the ROC curve [A(z)] = 0.92 +/- 0.01 [standard deviation] and (0.90)A(z) = 0.54 +/- 0.08 for LDA, A(z) = 0.92 +/- 0.01 and (0.90)A(z) = 0.55 +/- 0.08 for ANN). Results of both models generalized well to the retest set, with no significant performance differences between the validate and retest sets (P > .1). On the retest set, there were no significant performance differences between LDA with all features and LDA with only the stepwise-selected features (P > .3) and between either LDA and radiologist assessment score (P > .2). CONCLUSION: Results showed that combining mammographic and sonographic descriptors in a CAD model can result in high classification and generalization performance. On the retest set, LDA performance matched radiologist classification performance.

Authors
Jesneck, JL; Lo, JY; Baker, JA
MLA Citation
Jesneck, JL, Lo, JY, and Baker, JA. "Breast mass lesions: computer-aided diagnosis models with mammographic and sonographic descriptors." Radiology 244.2 (August 2007): 390-398.
PMID
17562812
Source
pubmed
Published In
Radiology
Volume
244
Issue
2
Publish Date
2007
Start Page
390
End Page
398
DOI
10.1148/radiol.2442060712

Incorporation of a Laguerre-Gauss channelized Hotelling observer for false-positive reduction in a mammographic mass CAD system.

Previously, we developed a simple Laguerre-Gauss (LG) channelized Hotelling observer (CHO) for incorporation into our mass computer-aided detection (CAD) system. This LG-CHO was trained using initial detection suspicious region data and was empirically optimized for free parameters. For the study presented in this paper, we wish to create a more optimal mass detection observer based on a novel combination of LG channels. A large set of LG channels with differing free parameters was created. Each of these channels was applied to the suspicious regions, and an output test statistic was determined. A stepwise feature selection algorithm was used to determine which LG channels would combine best to detect masses. These channels were combined using a HO to create a single template for the mass CAD system. Results from free-response receiver operating characteristic curves demonstrated that the incorporation of the novel LG-CHO into the CAD system slightly improved performance in high-sensitivity regions.

Authors
Baydush, AH; Catarious, DM; Lo, JY; Floyd, CE
MLA Citation
Baydush, AH, Catarious, DM, Lo, JY, and Floyd, CE. "Incorporation of a Laguerre-Gauss channelized Hotelling observer for false-positive reduction in a mammographic mass CAD system." J Digit Imaging 20.2 (June 2007): 196-202.
PMID
17505872
Source
pubmed
Published In
Journal of Digital Imaging
Volume
20
Issue
2
Publish Date
2007
Start Page
196
End Page
202
DOI
10.1007/s10278-007-9009-8

New developments in digital breast tomosynthesis

Authors
Lo, JY; Singh, S; III, DJT; Samei, E
MLA Citation
Lo, JY, Singh, S, III, DJT, and Samei, E. "New developments in digital breast tomosynthesis." June 2007.
Source
wos-lite
Published In
Medical physics
Volume
34
Issue
6
Publish Date
2007
Start Page
2518
End Page
2518
DOI
10.1118/1.2761222

Multiprojection correlation imaging for improved detection of pulmonary nodules.

OBJECTIVE: The purpose of this study was the development and preliminary evaluation of multiprojection correlation imaging with 3D computer-aided detection (CAD) on chest radiographs for cost- and dose-effective improvement of early detection of pulmonary nodules. SUBJECTS AND METHODS: Digital chest radiographs of 10 configurations of a chest phantom and of seven human subjects were acquired in multiple angular projections with an acquisition time of 11 seconds (single breath-hold) and total exposure comparable with that of a posteroanterior chest radiograph. An initial 2D CAD algorithm with two difference-of-gaussians filters and multilevel thresholds was developed with an independent database of 44 single-view chest radiographs with confirmed lesions. This 2D CAD algorithm was used on each projection image to find likely suspect nodules. The CAD outputs were reconstructed in 3D, reinforcing signals associated with true nodules while simultaneously decreasing false-positive findings produced by overlapping anatomic features. The performance of correlation imaging was tested on two to 15 projection images. RESULTS: Optimum performance of correlation imaging was attained when nine projection images were used. Compared with conventional, single-view CAD, correlation imaging decreased as much as 79% the frequency of false-positive findings in phantom cases at a sensitivity level of 65%. The corresponding reduction in false-positive findings in the cases of human subjects was 78%. CONCLUSION: Although limited by a relatively simple CAD implementation and a small number of cases, the findings suggest that correlation imaging performs substantially better than single-view CAD and may greatly enhance identification of subtle solitary pulmonary nodules on chest radiographs.

Authors
Samei, E; Stebbins, SA; Dobbins, JT; McAdams, HP; Lo, JY
MLA Citation
Samei, E, Stebbins, SA, Dobbins, JT, McAdams, HP, and Lo, JY. "Multiprojection correlation imaging for improved detection of pulmonary nodules." AJR Am J Roentgenol 188.5 (May 2007): 1239-1245.
PMID
17449766
Source
pubmed
Published In
AJR. American journal of roentgenology
Volume
188
Issue
5
Publish Date
2007
Start Page
1239
End Page
1245
DOI
10.2214/AJR.06.0843

Evaluation of information-theoretic similarity measures for content-based retrieval and detection of masses in mammograms.

The purpose of this study was to evaluate image similarity measures employed in an information-theoretic computer-assisted detection (IT-CAD) scheme. The scheme was developed for content-based retrieval and detection of masses in screening mammograms. The study is aimed toward an interactive clinical paradigm where physicians query the proposed IT-CAD scheme on mammographic locations that are either visually suspicious or indicated as suspicious by other cuing CAD systems. The IT-CAD scheme provides an evidence-based, second opinion for query mammographic locations using a knowledge database of mass and normal cases. In this study, eight entropy-based similarity measures were compared with respect to retrieval precision and detection accuracy using a database of 1820 mammographic regions of interest. The IT-CAD scheme was then validated on a separate database for false positive reduction of progressively more challenging visual cues generated by an existing, in-house mass detection system. The study showed that the image similarity measures fall into one of two categories; one category is better suited to the retrieval of semantically similar cases while the second is more effective with knowledge-based decisions regarding the presence of a true mass in the query location. In addition, the IT-CAD scheme yielded a substantial reduction in false-positive detections while maintaining high detection rate for malignant masses.

Authors
Tourassi, GD; Harrawood, B; Singh, S; Lo, JY; Floyd, CE
MLA Citation
Tourassi, GD, Harrawood, B, Singh, S, Lo, JY, and Floyd, CE. "Evaluation of information-theoretic similarity measures for content-based retrieval and detection of masses in mammograms." Med Phys 34.1 (January 2007): 140-150.
PMID
17278499
Source
pubmed
Published In
Medical physics
Volume
34
Issue
1
Publish Date
2007
Start Page
140
End Page
150
DOI
10.1118/1.2401667

Neutron stimulated emission computed tomography: Background corrections

Neutron stimulated emission computed tomography (NSECT) is an imaging technique that provides an in-vivo tomographic spectroscopic image of the distribution of elements in a body. To achieve this, a neutron beam illuminates the body. Nuclei in the body along the path of the beam are stimulated by inelastic scattering of the neutrons in the beam and emit characteristic gamma photons whose unique energy identifies the element. The emitted gammas are collected in a spectrometer and form a projection intensity for each spectral line at the projection orientation of the neutron beam. Rotating and translating either the body or the beam will allow a tomographic projection set to be acquired. Images are reconstructed to represent the spatial distribution of elements in the body. Critical to this process is the appropriate removal of background gamma events from the spectrum. Here we demonstrate the equivalence of two background correction techniques and discuss the appropriate application of each. © 2006 Elsevier B.V. All rights reserved.

Authors
Jr, CEF; Sharma, AC; Bender, JE; Kapadia, AJ; Xia, JQ; Harrawood, BP; Tourassi, GD; Lo, JY; Kiser, MR; Crowell, AS; Pedroni, RS; Macri, RA; Tajima, S; Howell, CR
MLA Citation
Jr, CEF, Sharma, AC, Bender, JE, Kapadia, AJ, Xia, JQ, Harrawood, BP, Tourassi, GD, Lo, JY, Kiser, MR, Crowell, AS, Pedroni, RS, Macri, RA, Tajima, S, and Howell, CR. "Neutron stimulated emission computed tomography: Background corrections." Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms 254.2 (2007): 329-336.
Source
scival
Published In
Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms
Volume
254
Issue
2
Publish Date
2007
Start Page
329
End Page
336
DOI
10.1016/j.nimb.2006.11.098

A comparison between traditional shift-and-add (SAA) and point-by-point back projection (BP) - Relevance to morphology of microcalcifications for isocentric motion in Digital Breast tomosynthesis (DBT)

Digital breast tomosynthesis (DBT) is a three-dimensional imaging technique providing an arbitrary set of reconstruction planes in the breast with limited series of projection images. This paper describes a comparison between traditional shift-and-add (SAA) and point-by-point back projection (BP) algorithms by impulse response and modulation transfer function (MTF) analysis. Due to the partial isocentric motion of the x-ray tube in DBT, structures such as microcalcifications appear slightly blurred in traditional shift-and-add (SAA) images in the direction perpendicular to the direction of tube 's motion. Point-by-point BP improved rendition of microcalcifications. The sharpness and morphology of calcifications were improved in a human subject images. A Filtered Back Projection (FBP) deblurring approach was used to demonstrate deblurred point-by-point BP tomosynthesis images. The point-by-point BP rather than traditional SAA should be considered as the foundation of further deblurring algorithms for DBT reconstruction. ©2007 IEEE.

Authors
Chen, Y; Lo, JY; III, JTD
MLA Citation
Chen, Y, Lo, JY, and III, JTD. "A comparison between traditional shift-and-add (SAA) and point-by-point back projection (BP) - Relevance to morphology of microcalcifications for isocentric motion in Digital Breast tomosynthesis (DBT)." Proceedings of the 7th IEEE International Conference on Bioinformatics and Bioengineering, BIBE (2007): 563-569.
Source
scival
Published In
Proceedings of the 7th IEEE International Conference on Bioinformatics and Bioengineering, BIBE
Publish Date
2007
Start Page
563
End Page
569
DOI
10.1109/BIBE.2007.4375617

Methodology of NEQ (f) analysis for optimization and comparison of digital breast tomosynthesis acquisition techniques and reconstruction algorithms

As a new three-dimensional imaging technique, digital breast tomosynthesis allows the reconstruction of an arbitrary set of planes in the breast from a limited-angle series of projection images. Though several tomosynthesis algorithms have been proposed, no complete optimization and comparison of different tomosynthesis acquisition techniques for available methods has been conducted as of yet. This paper represents a methodology of noise-equivalent quanta NEQ (f) analysis to optimize and compare the efficacy of tomosynthesis algorithms and imaging acquisition techniques for digital breast tomosynthesis. It combines the modulation transfer function (MTF) of system signal performance and the noise power spectrum (NPS) of noise characteristics. It enables one to evaluate the performance of different acquisition parameters and algorithms for comparison and optimization purposes. An example of this methodology was evaluated on a selenium-based direct-conversion flat-panel Siemens Mammomat Novation prototype system. An edge method was used to measure the presampled MTF of the detector. The MTF associated with the reconstruction algorithm and specific acquisition technique was investigated by calculating the Fourier Transform of simulated impulse responses. Flat field tomosynthesis projection sequences were acquired and then reconstructed. A mean-subtracted NPS on the reconstructed plane was studied to remove fixed pattern noise. An example of the application of this methodology was illustrated in this paper using a point-by-point Back Projection correction (BP) reconstruction algorithm and an acquisition technique of 25 projections with 25 degrees total angular tube movement.

Authors
Chen, Y; Lo, JY; Ranger, NT; Samei, E; III, JTD
MLA Citation
Chen, Y, Lo, JY, Ranger, NT, Samei, E, and III, JTD. "Methodology of NEQ (f) analysis for optimization and comparison of digital breast tomosynthesis acquisition techniques and reconstruction algorithms." 2007.
Source
scival
Published In
Proceedings of SPIE
Volume
6510
Issue
PART 1
Publish Date
2007
DOI
10.1117/12.713737

Efficient registration of aliased x-ray images

Multiframe image reconstruction produces images beyond the native resolution of a digital image sensor by way of accurate sub-pixel registration of aliased images. We present a novel multiframe registration approach for the purpose of enhancing resolution of digital mammogram images. We demonstrate the ability to improve resolution while maintaining normal radiation dosages. © 2007 IEEE.

Authors
Robinson, MD; Farsiu, S; Lo, JY; Milanfar, P; Toth, CA
MLA Citation
Robinson, MD, Farsiu, S, Lo, JY, Milanfar, P, and Toth, CA. "Efficient registration of aliased x-ray images." Conference Record - Asilomar Conference on Signals, Systems and Computers (2007): 215-219.
Source
scival
Published In
Conference Record of the Asilomar Conference on Signals, Systems and Computers
Publish Date
2007
Start Page
215
End Page
219
DOI
10.1109/ACSSC.2007.4487198

Decision fusion of circulating markers for breast cancer detection in premenopausal women

Current mammographic screeningfor breast cancer is less effective for younger women. To complement mammography for premenopausal women, we investigated the feasibility screening test using 98 blood serum proteins. Because the data set was very noisy and contained only weak features, we used a classifier designed for noisy data: decision fusion. Decision fusion outperformed both a support vector machine (SVM) and linear regression with forward stepwise feature selection on all three two-class classification tasks: normal tissue vs. cancer, normal tissue vs. benign lesions, and benign lesions vs. cancer. Decision fusion detected cancer moderately well (AUC=0.84 on normal vs. cancer), demonstrating promise as a screening tool. Decision fusion also detected benign lesions similarly well (AUC=0.83 on normal vs. benign lesions) and was the only classifier to achieve any success in separating benign from malignant lesions (AUC=0.64 on benign vs. cancer). The classification results suggest that the assayed proteins are more indicative of a secondary effect, such as immune response, rather than specific for breast cancer. In conclusion, the decision fusion classifier demonstrated some promise in detecting breast lesions and outperformed other classifiers, especially for the very noisy classification problem of distinguishing benign from malignant lesions. ©2007 IEEE.

Authors
Jesneck, JL; Mukherjee, S; Nolte, LW; Lokshin, AE; Marks, JR; Lo, J
MLA Citation
Jesneck, JL, Mukherjee, S, Nolte, LW, Lokshin, AE, Marks, JR, and Lo, J. "Decision fusion of circulating markers for breast cancer detection in premenopausal women." 2007. 1434-1438.
Source
scival
Publish Date
2007
Start Page
1434
End Page
1438
DOI
10.1109/BIBE.2007.4375762

Feasibility study of breast tomosynthesis CAD system

The purpose of this study was to investigate feasibility of computer-aided detection of masses and calcification clusters in breast tomosynthesis images and obtain reliable estimates of sensitivity and false positive rate on an independent test set. Automatic mass and calcification detection algorithms developed for film and digital mammography images were applied without any adaptation or retraining to tomosynthesis projection images. Test set contained 36 patients including 16 patients with 20 known malignant lesions, 4 of which were missed by the radiologists in conventional mammography images and found only in retrospect in tomosynthesis. Median filter was applied to tomosynthesis projection images. Detection algorithm yielded 80% sensitivity and 5.3 false positives per breast for calcification and mass detection algorithms combined. Out of 4 masses missed by radiologists in conventional mammography images, 2 were found by the mass detection algorithm in tomosynthesis images.

Authors
Jerebko, A; Quan, Y; Merlet, N; Ratner, E; Singh, S; Lo, JY; Krishnan, A
MLA Citation
Jerebko, A, Quan, Y, Merlet, N, Ratner, E, Singh, S, Lo, JY, and Krishnan, A. "Feasibility study of breast tomosynthesis CAD system." 2007.
Source
scival
Published In
Proceedings of SPIE
Volume
6514
Issue
PART 1
Publish Date
2007
DOI
10.1117/12.712729

Initial human subject results for breast Bi-plane correlation imaging technique

Computer aided detection (CADe) systems often present multiple false-positives per image in projection mammography due to overlapping anatomy. To reduce the number of such false-positives, we propose performing CADe on image pairs acquired using a bi-plane correlation imaging (BCI) technique. In this technique, images are acquired of each breast at two different projection angles. A traditional CADe algorithm operates on each image to identify suspected lesions. The suspicious areas from both projections are then geometrically correlated, eliminating any lesion that is not identified on both views. Proof of concept studies showed that that the BCI technique reduced the numbers of false-positives per case up to 70%. (This work was supported in part by grants from the Department of Defense (USAMRMC W81XWH-04-1-0323), Komen Foundation (PDF55806), NIH (R01-CA-109074 and R01-CA-112437), Cancer Research and Prevention Foundation, and a research agreement with Siemens Medical Solutions.).

Authors
Jr, RSS; Samei, E; Majdi-Nasab, N; Lo, JY
MLA Citation
Jr, RSS, Samei, E, Majdi-Nasab, N, and Lo, JY. "Initial human subject results for breast Bi-plane correlation imaging technique." 2007.
Source
scival
Published In
Proceedings of SPIE
Volume
6514
Issue
PART 2
Publish Date
2007
DOI
10.1117/12.713722

Breast mass detection in tomosynthesis projection images using information-theoretic similarity measures

The purpose of this project is to study Computer Aided Detection (CADe) of breast masses for digital tomosynthesis. It is believed that tomosynthesis will show improvement over conventional mammography in detection and characterization of breast masses by removing overlapping dense fibroglandular tissue. This study used the 60 human subject cases collected as part of on-going clinical trials at Duke University. Raw projections images were used to identify suspicious regions in the algorithm's high-sensitivity, low-specificity stage using a Difference of Gaussian (DoG) filter. The filtered images were thresholded to yield initial CADe hits that were then shifted and added to yield a 3D distribution of suspicious regions. These were further summed in the depth direction to yield a flattened probability map of suspicious hits for ease of scoring. To reduce false positives, we developed an algorithm based on information theory where similarity metrics were calculated using knowledge databases consisting of tomosynthesis regions of interest (ROIs) obtained from projection images. We evaluated 5 similarity metrics to test the false positive reduction performance of our algorithm, specifically joint entropy, mutual information, Jensen difference divergence, symmetric Kullback-Liebler divergence, and conditional entropy. The best performance was achieved using the joint entropy similarity metric, resulting in ROC A 2 of 0.87 ±0.01. As a whole, the CADe system can detect breast masses in this data set with 79% sensitivity and 6.8 false positives per scan. In comparison, the original radiologists performed with only 65% sensitivity when using mammography alone, and 91% sensitivity when using tomosynthesis alone.

Authors
Singh, S; Tourassi, GD; Lo, JY
MLA Citation
Singh, S, Tourassi, GD, and Lo, JY. "Breast mass detection in tomosynthesis projection images using information-theoretic similarity measures." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6514.PART 1 (2007).
Source
scival
Published In
Proceedings of SPIE
Volume
6514
Issue
PART 1
Publish Date
2007
DOI
10.1117/12.713004

Visual image quality metrics for optimization of breast tomosynthesis acquisition technique

Breast tomosynthesis is currently an investigational imaging technique requiring optimization of its many combinations of data acquisition and image reconstruction parameters for optimum clinical use. In this study, the effects of several acquisition parameters on the visual conspicuity of diagnostic features were evaluated for three breast specimens using a visual discrimination model (VDM). Acquisition parameters included total exposure, number of views, full resolution and binning modes, and lag correction. The diagnostic features considered in these specimens were mass margins, microcalcifications, and mass spicules. Metrics of feature contrast were computed for each image by defining two regions containing the selected feature (Signal) and surrounding background (Noise), and then computing the difference in VDM channel metrics between Signal and Noise regions in units of just-noticeable differences (JNDs). Scans with 25 views and exposure levels comparable to a standard two-view mammography exam produced higher levels of feature contrast. The effects of binning and lag correction on feature contrast were found to be generally small and isolated, consistent with our visual assessments of the images. Binning produced a slight loss of spatial resolution which could be compensated in the reconstruction filter. These results suggest that good image quality can be achieved with the faster and therefore more clinically practical 25-view scans with binning, which can be performed in as little as 12.5 seconds. Further work will investigate other specimens as well as alternate figures of merit in order to help determine optimal acquisition and reconstruction parameters for clinical trials.

Authors
Johnson, JP; Lo, J; Mertelmeier, T; Nafziger, JS; Timberg, P; Samei, E
MLA Citation
Johnson, JP, Lo, J, Mertelmeier, T, Nafziger, JS, Timberg, P, and Samei, E. "Visual image quality metrics for optimization of breast tomosynthesis acquisition technique." 2007.
Source
scival
Published In
Proceedings of SPIE
Volume
6515
Publish Date
2007
DOI
10.1117/12.712343

On the development of a Gaussian noise model for scatter compensation

The underlying mechanism in projection radiography as well as in computed tomography (CT) is the accumulative attenuation of a pencil x-ray beam along a straight line. However, when a portion of photons is deviated from their original path by scattering, it is not valid to assume that these photons are the survival photons along the lines connecting the x-ray source and the individual locations where they are detected. Since these photons do not carry the correct spatial information, the final image is contaminated. Researchers are seeking techniques to reduce scattering, and hence, improve image quality, by scatter compensation. Previously, we presented a post-acquisition scatter compensation technique based on an underlying statistical model. We used the Poisson noise model, which assumed that the signals in the detector individually followed the Poisson process. Since most x-ray detectors are energy integrating rather than photon counting, the Poisson noise model can be improved by taking this property into account. In this study, we developed a Gaussian noise model by the matching-of-the-first-two-moments method. The Maximum Likelihood Estimator of the scatter-free image was derived via the expectation maximization (EM) technique. The maximum a posteriori estimate was also calculated. The Gaussian noise model was preliminarily evaluated on a full-field digital mammography system.

Authors
Xia, JQ; Tourassi, GD; Lo, JY; Jr, CEF
MLA Citation
Xia, JQ, Tourassi, GD, Lo, JY, and Jr, CEF. "On the development of a Gaussian noise model for scatter compensation." 2007.
Source
scival
Published In
Proceedings of SPIE
Volume
6510
Issue
PART 2
Publish Date
2007
DOI
10.1117/12.712515

Breast mass detection in tomosynthesis projection images using information-theoretic similarity measures - art. no. 651415

Authors
Singh, S; Tourassi, GD; Lo, JY
MLA Citation
Singh, S, Tourassi, GD, and Lo, JY. "Breast mass detection in tomosynthesis projection images using information-theoretic similarity measures - art. no. 651415." Medical Imaging 2007: Computer-Aided Diagnosis, Pts 1 and 2 6514 (2007): 51415-51415.
Source
wos-lite
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
6514
Publish Date
2007
Start Page
51415
End Page
51415
DOI
10.1117/12.713044

Optimized approach to decision fusion of heterogeneous data for breast cancer diagnosis.

As more diagnostic testing options become available to physicians, it becomes more difficult to combine various types of medical information together in order to optimize the overall diagnosis. To improve diagnostic performance, here we introduce an approach to optimize a decision-fusion technique to combine heterogeneous information, such as from different modalities, feature categories, or institutions. For classifier comparison we used two performance metrics: The receiving operator characteristic (ROC) area under the curve [area under the ROC curve (AUC)] and the normalized partial area under the curve (pAUC). This study used four classifiers: Linear discriminant analysis (LDA), artificial neural network (ANN), and two variants of our decision-fusion technique, AUC-optimized (DF-A) and pAUC-optimized (DF-P) decision fusion. We applied each of these classifiers with 100-fold cross-validation to two heterogeneous breast cancer data sets: One of mass lesion features and a much more challenging one of microcalcification lesion features. For the calcification data set, DF-A outperformed the other classifiers in terms of AUC (p < 0.02) and achieved AUC=0.85 +/- 0.01. The DF-P surpassed the other classifiers in terms of pAUC (p < 0.01) and reached pAUC=0.38 +/- 0.02. For the mass data set, DF-A outperformed both the ANN and the LDA (p < 0.04) and achieved AUC=0.94 +/- 0.01. Although for this data set there were no statistically significant differences among the classifiers' pAUC values (pAUC=0.57 +/- 0.07 to 0.67 +/- 0.05, p > 0.10), the DF-P did significantly improve specificity versus the LDA at both 98% and 100% sensitivity (p < 0.04). In conclusion, decision fusion directly optimized clinically significant performance measures, such as AUC and pAUC, and sometimes outperformed two well-known machine-learning techniques when applied to two different breast cancer data sets.

Authors
Jesneck, JL; Nolte, LW; Baker, JA; Floyd, CE; Lo, JY
MLA Citation
Jesneck, JL, Nolte, LW, Baker, JA, Floyd, CE, and Lo, JY. "Optimized approach to decision fusion of heterogeneous data for breast cancer diagnosis." Med Phys 33.8 (August 2006): 2945-2954.
Website
http://hdl.handle.net/10161/207
PMID
16964873
Source
pubmed
Published In
Medical physics
Volume
33
Issue
8
Publish Date
2006
Start Page
2945
End Page
2954
DOI
10.1118/1.2208934

Introduction to neutron stimulated emission computed tomography.

Neutron stimulated emission computed tomography (NSECT) is presented as a new technique for in vivo tomographic spectroscopic imaging. A full implementation of NSECT is intended to provide an elemental spectrum of the body or part of the body being interrogated at each voxel of a three-dimensional computed tomographic image. An external neutron beam illuminates the sample and some of these neutrons scatter inelastically, producing characteristic gamma emission from the scattering nuclei. These characteristic gamma rays are acquired by a gamma spectrometer and the emitting nucleus is identified by the emitted gamma energy. The neutron beam is scanned over the body in a geometry that allows for tomographic reconstruction. Tomographic images of each element in the spectrum can be reconstructed to represent the spatial distribution of elements within the sample. Here we offer proof of concept for the NSECT method, present the first single projection spectra acquired from multi-element phantoms, and discuss potential biomedical applications.

Authors
Floyd, CE; Bender, JE; Sharma, AC; Kapadia, A; Xia, J; Harrawood, B; Tourassi, GD; Lo, JY; Crowell, A; Howell, C
MLA Citation
Floyd, CE, Bender, JE, Sharma, AC, Kapadia, A, Xia, J, Harrawood, B, Tourassi, GD, Lo, JY, Crowell, A, and Howell, C. "Introduction to neutron stimulated emission computed tomography." Phys Med Biol 51.14 (July 21, 2006): 3375-3390.
PMID
16825736
Source
pubmed
Published In
Physics in Medicine and Biology
Volume
51
Issue
14
Publish Date
2006
Start Page
3375
End Page
3390
DOI
10.1088/0031-9155/51/14/006

Computer-aided diagnosis in breast imaging: Where do we go after detection?

Authors
Lo, JY; Bilska-Wolak, AO; Baker, JA; Tourassi, GD; Floyd, CE; Markey, MK
MLA Citation
Lo, JY, Bilska-Wolak, AO, Baker, JA, Tourassi, GD, Floyd, CE, and Markey, MK. "Computer-aided diagnosis in breast imaging: Where do we go after detection?." Recent Advances in Breast Imaging, Mammography, and Computer-Aided Diagnosis of Breast Cancer. April 10, 2006. 871-900.
Source
scopus
Publish Date
2006
Start Page
871
End Page
900
DOI
10.1117/3.651880.ch27

Using computational intelligence for computer-aided diagnosis of screen-film mammograms

Authors
Land, WH; McKee, DW; Anderson, FR; Masters, T; Lo, JY; Embrechts, M; Heine, J
MLA Citation
Land, WH, McKee, DW, Anderson, FR, Masters, T, Lo, JY, Embrechts, M, and Heine, J. "Using computational intelligence for computer-aided diagnosis of screen-film mammograms." Recent Advances in Breast Imaging, Mammography, and Computer-Aided Diagnosis of Breast Cancer. April 10, 2006. 315-375.
Source
scopus
Publish Date
2006
Start Page
315
End Page
375
DOI
10.1117/3.651880.ch10

Noise power spectrum analysis for several digital breast tomosynthesis reconstruction algorithms

Digital breast tomosynthesis is a three-dimensional imaging technique that allows the reconstruction of an arbitrary set of planes in the breast from limited-angle series of projection images. Though several tomosynthesis algorithms have been proposed, no complete optimization and comparison of all available methods has been conducted as of yet. This paper presents an analysis of noise power spectrum to examine the noise characteristics of several tomosynthesis algorithms with different imaging acquisition techniques. Flat images were acquired with the following acquisition parameters: 13, 25, 49 projections with ±12.5 and ±25 degrees of angular ranges. Three algorithms, including Shift-And-Add (SAA), Matrix Inversion Tomosynthesis (MITS), and Filtered Back Projection (FBP) were investigated with reconstruction slice spacing of 1mm, 2mm, and 4mm. The noise power spectra of the reconstruction plane at 23.5mm above the detector surface were analyzed. Results showed that MITS has better noise responses with narrower slice spacing for low-to-middle frequencies. No substantial difference was noticed for SAA and FBP with different slice spacings. With the same acquisition technique and slice spacing, MITS performed better than FBP at middle frequencies, but FBP showed better performance at high frequencies because of applied Hamming and Gaussian low-pass filters. For different imaging acquisition techniques, SAA, MITS and FBP performed the best with 49 projections and ±25 degrees. For 25 projections specifically, FBP performed better with wider angular range, while MITS performed better with narrower angular range. For SAA, narrow angular range is slightly better for 25 projections and 13 projections.

Authors
Chen, Y; Lo, JY; III, JTD
MLA Citation
Chen, Y, Lo, JY, and III, JTD. "Noise power spectrum analysis for several digital breast tomosynthesis reconstruction algorithms." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6142 III (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6142 III
Publish Date
2006
DOI
10.1117/12.652282

Gaussian frequency blending algorithm with Matrix Inversion Tomosynthesis (MITS) and Filtered Back Projection (FBP) for better digital breast tomosynthesis reconstruction

Breast cancer is a major problem and the most common cancer among women. The nature of conventional mammography makes it very difficult to distinguish a cancer from overlying breast tissues. Digital Tomosynthesis refers to a three-dimensional imaging technique that allows reconstruction of an arbitrary set of planes in the breast from limited-angle series of projection images as the x-ray source moves. Several tomosynthesis algorithms have been proposed, including Matrix Inversion Tomosynthesis (MITS) and Filtered Back Projection (FBP) that have been investigated in our lab. MITS shows better high frequency response in removing out-of-plane blur, while FBP shows better low frequency noise prosperities. This paper presents an effort to combine MITS and FBP for better breast tomosynthesis reconstruction. A high-pass Gaussian filter was designed and applied to three-slice "slabbing" MITS reconstructions. A low-pass Gaussian filter was designed and applied to the FBP reconstructions. A frequency weighting parameter was studied to blend the high-passed MITS with low-passed FBP frequency components. Four different reconstruction methods were investigated and compared with human subject images: 1) MITS blended with Shift-And-Add (SAA), 2) FBP alone, 3) FBP with applied Hamming and Gaussian Filters, and 4) Gaussian Frequency Blending (GFB) of MITS and FBP. Results showed that, compared with FBP, Gaussian Frequency Blending (GFB) has better performance for high frequency content such as better reconstruction of micro-calcifications and removal of high frequency noise. Compared with MITS, GFB showed more low frequency breast tissue content.

Authors
Chen, Y; Lo, JY; Baker, JA; III, JTD
MLA Citation
Chen, Y, Lo, JY, Baker, JA, and III, JTD. "Gaussian frequency blending algorithm with Matrix Inversion Tomosynthesis (MITS) and Filtered Back Projection (FBP) for better digital breast tomosynthesis reconstruction." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6142 I (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6142 I
Publish Date
2006
DOI
10.1117/12.652264

Breast cancer diagnosis using neutron stimulated emission computed tomography: Dose and count requirements

Neutron Stimulated Emission Computed Tomography (NSECT) was evaluated as a potential technique for breast cancer diagnosis. NSECT can form a 3D tomographic image with an elemental (isotopic) spectrum provided at each reconstructed voxel. The target is illuminated (in vivo) by a neutron beam that scatters in-elastically producing characteristic gamma emission that is acquired tomographically with a spectrograph. Images are reconstructed of each element in the acquired spectrum. NSECT imaging was simulated for benign and malignant breast masses. A range of the number of incident neutrons was simulated from 19 million to 500k neutrons. Simulation included all known primary and secondary physical interactions in both the breast as well as in the spectrometer. Characteristic energy spectra were acquired by simulation and were analyzed for statistically significant differences between benign and malignant breasts. For 1 million incident neutrons, there were 61 differences in the spectra that were statistically significant (p < 0,05), Of these, 23 matched known characteristic emission from 6 elements that have been found in the breast (Br, Cs, K, Mn, Rb, Zn). The dose to two breasts was less than 3% of the dose of a 4 view screening mammogram, Increasing the dose to 52% of the mammogram (19 million neutrons) provided 89 significant spectral differences that matched 30 known emissions from 7 elements that have been found in the breast (Br, Co, Cs, K, Mn, Rb, Zn). Decreasing the dose to 1.4% (500K neutrons) eliminated all statistically significant matches to known elements. This study suggests that NSECT may be a viable technique for detecting human breast cancer in vivo at a reduced dose compared to 4 view screening mammography.

Authors
Jr, CEF; Bender, JE; Harrawood, B; Sharma, AC; Kapadia, A; Tourassi, GD; Lo, JY; Howell, C
MLA Citation
Jr, CEF, Bender, JE, Harrawood, B, Sharma, AC, Kapadia, A, Tourassi, GD, Lo, JY, and Howell, C. "Breast cancer diagnosis using neutron stimulated emission computed tomography: Dose and count requirements." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6142 II (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6142 II
Publish Date
2006
DOI
10.1117/12.656045

The effect of data set size on computer-aided diagnosis of breast cancer: Comparing decision fusion to a linear discriminant

Data sets with relatively few observations (cases) in medical research are common, especially if the data are expensive or difficult to collect. Such small sample sizes usually do not provide enough information for computer models to learn data patterns well enough for good prediction and generalization. As a model that may be able to maintain good classification performance in the presence of limited data, we used decision fusion. In this study, we investigated the effect of sample size on the generalization ability of both linear discriminant analysis (LDA) and decision fusion. Subsets of large data sets were selected by a bootstrap sampling method, which allowed us to estimate the mean and standard deviation of the classification performance as a function of data set size. We applied the models to two breast cancer data sets and compared the models using receiver operating characteristic (ROC) analysis. For the more challenging calcification data set, decision fusion reached its maximum classification performance of AUC = 0.80±0.04 at 50 samples and pAUC = 0.34±0.05 at 100 samples. The LDA reached a lower performance and required many more cases, with a maximum of AUC = 0.68±0.04 and pAUC = 0.12±0.05 at 450 samples. For the mass data set, the two classifiers had more similar performance, with AUC = 0.92±0.02 and pAUC = 0.48±0.02 at 50 samples for decision fusion and AUC = 0.92±0.03 and pAUC = 0.55±0.04 at 500 samples for the LDA.

Authors
Jesneck, JL; Nolte, LW; Baker, JA; Lo, JY
MLA Citation
Jesneck, JL, Nolte, LW, Baker, JA, and Lo, JY. "The effect of data set size on computer-aided diagnosis of breast cancer: Comparing decision fusion to a linear discriminant." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6146 (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6146
Publish Date
2006
DOI
10.1117/12.655235

Rotating slat collimator design for high-energy near-field imaging

Certain elements (such as Fe, Cu, Zn, etc.) are vital to the body and an imbalance of such elements can either be a symptom or cause of certain pathologies. Neutron Stimulated Emission Computed Tomography (NSECT) is a spectroscopic imaging technique whereby the body is illuminated via a beam of neutrons causing elemental nuclei to become excited and emit characteristic gamma radiation. Acquiring the gamma energy spectra in a tomographic geometry allows reconstruction of elemental concentration images. Previously we have demonstrated the feasibility of NSECT using first generation CT approaches; while successful, the approach does not scale well and has limited resolution. Additionally, current gamma cameras operate in an energy range too low for NSECT imaging. However, the orbiting Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI) captures and images gamma rays over the high-energy range equivalent to NSECT's (3 keV to 17 MeV) by utilizing Collimator-based Fourier transform imaging. A High Purity Germanium (HPGe) detector counts the number of energy events per unit of time, providing spectroscopic data. While a pair of rotating collimators placed in front of the detector modulates the number of gamma events, providing spatial information. Knowledge of the number of energy events at each discrete collimator angle allows for 2D image reconstruction. This method has proven successful at a focus of infinity in the RHESSI application. Our goal is to achieve similar results at a reasonable near-field focus. Here we describe the results of our simulations to implement a rotating modulation collimator (RMC) gamma imager for use in NSECT using simulations in Matlab. To determine feasible collimator setups and the stability of the inverse problem a Matlab environment was created that uses the geometry of the system to generate ID observation data from 2D images and then to reconstruct 2D images using the MLEM algorithm. Reasonable collimator geometries were determined, successful reconstruction was achieved and the inverse problem was found to be stable.

Authors
Sharma, A; Floyd, C; Harrawood, B; Tourassi, G; Kapadia, A; Bender, J; Lo, J; Howell, C
MLA Citation
Sharma, A, Floyd, C, Harrawood, B, Tourassi, G, Kapadia, A, Bender, J, Lo, J, and Howell, C. "Rotating slat collimator design for high-energy near-field imaging." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6142 I (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6142 I
Publish Date
2006
DOI
10.1117/12.653929

Beam optimization for digital mammography - II

Optimization of acquisition technique factors (target, filter, and kVp) in digital mammography is required for maximization of the image SNR, while minimizing patient dose. The goal of this study is to compare, for each of the major commercially available FFDM systems, the effect of various technique factors on image SNR and radiation dose for a range of breast thickness and tissue types. This phantom study follows the approach of an earlier investigation[1], and includes measurements on recent versions of two of the FFDM systems discussed in that paper, as well as on three FFDM systems not available at that time, The five commercial FFDM systems tested are located at five different university test sites and include all FFDM systems that are currently FDA approved. Performance was assessed using 9 different phantom types (three compressed thicknesses, and three tissue composition types) using all available x-ray target and filter combinations, The figure of merit (FOM) used to compare technique factors is the ratio of the square of the image SNR to the mean glandular dose (MGD). This FOM has been used previously by others in mammographic beam optimization studies [2],[3]. For selected examples, data are presented describing the change in SNR, MOD, and FOM with changing kVp, as well as with changing target and/or filter type. For all nine breast types the target/filter/kVp combination resulting in the highest FOM value is presented. Our results suggest that in general, technique combinations resulting in higher energy beams resulted in higher FOM values, for nearly all breast types. © Springer-Verlag Berlin Heidelberg 2006.

Authors
Williams, MB; Raghunathan, P; Seibert, A; Kwan, A; Lo, J; Samei, E; Fajardo, L; Maidment, ADA; Yaffe, M; Bloomquist, A
MLA Citation
Williams, MB, Raghunathan, P, Seibert, A, Kwan, A, Lo, J, Samei, E, Fajardo, L, Maidment, ADA, Yaffe, M, and Bloomquist, A. "Beam optimization for digital mammography - II." Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 4046 LNCS (2006): 273-280.
Source
scival
Published In
Lecture notes in computer science
Volume
4046 LNCS
Publish Date
2006
Start Page
273
End Page
280

Mass detection in mammographic ROIs using Watson filters

Human vision models have been shown to capture the response of the visual system; their incorporation into the classification stage of a Computer Aided Detection system could improve performance. This study seeks to improve the performance of an automated breast mass detection system by using the Watson filter model versus a Laguerre Gauss Channelized Hotelling Observer (LG-CHO). The LG-CHO and the Watson filter model were trained and tested on a 512×512 ROI database acquired from the Digital Database of Screening Mammography consisting of 800 total ROIs; 200 of which were malignant, 200 were benign and 400 were normal. Half of the ROIs were used to train the weights for ten LG-CHO templates that were later used during the testing stage. For the Watson filter model, the training cases were used to optimize the frequency filter parameter empirically to yield the best ROC Az performance. This set of filter parameters was then tested on the remaining cases. The training Az for the LG-CHO and the Watson filter was 0.896 +/- 0.016 and 0.924 +/- 0.014 respectively. The testing Az for the LG-CHO and Watson filter was 0.849 +/- 0.019 and 0.888 +/- 0.017. With a p-value of 0.029, the difference in testing performance was statistically significant, thus implying that the Watson filter model holds promise for better detection of masses.

Authors
Singh, S; Baydush, A; Harrawood, B; Lo, J
MLA Citation
Singh, S, Baydush, A, Harrawood, B, and Lo, J. "Mass detection in mammographic ROIs using Watson filters." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 6146 (2006).
Source
scival
Published In
Proceedings of SPIE
Volume
6146
Publish Date
2006
DOI
10.1117/12.653224

Computer aid for decision to biopsy breast masses on mammography: validation on new cases.

RATIONALE AND OBJECTIVES: The purpose of this study was to validate the performance of a previously developed computer aid for breast mass classification for mammography on a new, independent database of cases not used for algorithm development. MATERIALS AND METHODS: A computer aid (classifier) based on the likelihood ratio (LRb) was previously developed on a database of 670 mass cases. The 670 cases (245 malignant) from one medical institution were described using 16 features from the American College of Radiology Breast Imaging-Reporting and Data System lexicon and patient history findings. A separate database of 151 (43 malignant) validation cases were collected that were previously unseen by the classifier. These new validation cases were evaluated by the classifier without retraining. Performance evaluation methods included Receiver Operating Characteristic (ROC), round-robin, and leave-one-out bootstrap sampling. RESULTS: The performance of the classifier on the training data yielded an average ROC area of 0.90 +/- 0.02 and partial ROC area (0.90AUC) of 0.60 +/- 0.06. The exact nonparametric performance on the validation set of 151 cases yielded a ROC area of 0.88 and 0.90AUC of 0.57. Using a 100% sensitivity cutoff threshold established on the training data (100% negative predictive value), the classifier correctly identified 100% of the malignant masses in the validation test set, while potentially obviating 26% of the biopsies performed on benign masses. CONCLUSION: The LRb classifier performed consistently on new data that was not used for classifier development. The LRb classifier shows promise as a potential aid in reducing the number of biopsies performed on benign masses.

Authors
Bilska-Wolak, AO; Floyd, CE; Lo, JY; Baker, JA
MLA Citation
Bilska-Wolak, AO, Floyd, CE, Lo, JY, and Baker, JA. "Computer aid for decision to biopsy breast masses on mammography: validation on new cases." Acad Radiol 12.6 (June 2005): 671-680.
PMID
15935965
Source
pubmed
Published In
Academic Radiology
Volume
12
Issue
6
Publish Date
2005
Start Page
671
End Page
680
DOI
10.1016/j.acra.2005.02.011

Comparative scatter and dose performance of slot-scan and full-field digital chest radiography systems.

PURPOSE: To evaluate the scatter, dose, and effective detective quantum efficiency (DQE) performance of a slot-scan digital chest radiography system compared with that of a full-field digital radiography system. MATERIALS AND METHODS: Scatter fraction of a slot-scan system was measured for an anthropomorphic and a geometric phantom by using a posterior beam-stop technique at 117 and 140 kVp. Measurements were repeated with a full-field digital radiography system with and without a 13:1 antiscatter grid at 120 and 140 kVp. For both systems, the effective dose was measured on posteroanterior and lateral views for standard clinical techniques by using dosimeters embedded in a female phantom. The effective DQEs of the two systems were assessed by taking into account the scatter performance and the DQE of each system. The statistical significance of all the comparative differences was ascertained by means of t test analysis. RESULTS: The slot-scan system and the full-field system with grid yielded scatter fractions of 0.13-0.14 and 0.42-0.48 in the lungs and 0.30-0.43 and 0.69-0.78 in the mediastinum, respectively. The sum of the effective doses for posteroanterior and lateral views for the slot-scan system (0.057 mSv +/- 0.003 [+/- standard deviation]) was 34% lower than that for the full-field system (0.086 mSv +/- 0.001, P < .05) at their respective clinical peak voltages (140 and 120 kVp, respectively). The effective DQE of the slot-scan system was equivalent to that of the full-field system in the lung region but was 37% higher in the dense regions (P < .05). CONCLUSION: The slot-scan design leads to marked scatter reduction compared with the more conventional full-field geometries with a grid. The improved scatter performance of a slot-scan geometry can effectively compensate for low DQE and lead to improved image quality.

Authors
Samei, E; Lo, JY; Yoshizumi, TT; Jesneck, JL; Dobbins, JT; Floyd, CE; McAdams, HP; Ravin, CE
MLA Citation
Samei, E, Lo, JY, Yoshizumi, TT, Jesneck, JL, Dobbins, JT, Floyd, CE, McAdams, HP, and Ravin, CE. "Comparative scatter and dose performance of slot-scan and full-field digital chest radiography systems." Radiology 235.3 (June 2005): 940-949.
PMID
15845791
Source
pubmed
Published In
Radiology
Volume
235
Issue
3
Publish Date
2005
Start Page
940
End Page
949
DOI
10.1148/radiol.2353040516

Accuracy of segmentation of a commercial computer-aided detection system for mammography.

PURPOSE: To assess the accuracy of segmentation in a commercially available computer-aided detection (CAD) system. MATERIALS AND METHODS: Approval for this study was obtained from the authors' institutional review board. Informed consent was not required by the board for this review, as data were stripped of patient identifiers. Two thousand twenty mammograms from 507 women were analyzed with the hardware and software of a commercial CAD system. The accuracy of the segmentation process was determined semiquantitatively and categorized as near perfect if the skin line of the breast was accurately detected, acceptable if only subcutaneous fat was excluded, or unacceptable if any breast parenchyma was excluded from consideration. The accuracy of segmentation was compared for different breast densities and film sizes by using logistic regression (P < .05). RESULTS: Overall, segmentation was near perfect or acceptable in almost 96.8% of images. However, segmentation defects were significantly more common in mammograms with heterogeneously dense breast tissue (8% unacceptable) than in those with fatty replaced (0% unacceptable), scattered (1.2% unacceptable), or extremely dense (1.8% unacceptable) breast parenchyma (P < .05). For images with unacceptable segmentation, the average percentage of breast parenchyma excluded was almost 25% (range, 5%-100%), with no significant differences among breast densities. CONCLUSION: For one commercial CAD system, segmentation was usually near perfect or acceptable but was unacceptable more than five times more frequently for mammograms of breasts with heterogeneously dense parenchyma than for those with all other breast densities. On average, one-quarter of the breast parenchyma was excluded from CAD analysis for images with unacceptable segmentation.

Authors
Baker, JA; Rosen, EL; Crockett, MM; Lo, JY
MLA Citation
Baker, JA, Rosen, EL, Crockett, MM, and Lo, JY. "Accuracy of segmentation of a commercial computer-aided detection system for mammography." Radiology 235.2 (May 2005): 385-390.
PMID
15770043
Source
pubmed
Published In
Radiology
Volume
235
Issue
2
Publish Date
2005
Start Page
385
End Page
390
DOI
10.1148/radiol.2352040899

Physical characterization of a prototype selenium-based full field digital mammography detector.

The purpose of this study was to measure experimentally the physical performance of a prototype mammographic imager based on a direct detection, flat-panel array design employing an amorphous selenium converter with 70 microm pixels. The system was characterized for two different anode types, a molybdenum target with molybdenum filtration (Mo/Mo) and a tungsten target with rhodium filtration (W/Rh), at two different energies, 28 and 35 kVp, with approximately 2 mm added aluminum filtration. To measure the resolution, the presampled modulation transfer function (MTF) was measured using an edge method. The normalized noise power spectrum (NNPS) was measured by two-dimensional Fourier analysis of uniformly exposed mammograms. The detective quantum efficiencies (DQEs) were computed from the MTFs, the NNPSs, and theoretical ideal signal to noise ratios. The MTF was found to be close to its ideal limit and reached 0.2 at 11.8 mm(-1) and 0.1 at 14.1 mm(-1) for images acquired at an RQA-M2 technique (Mo/Mo anode, 28 kVp, 2 mm Al). Using a tungsten technique (MW2; W/Rh anode, 28 kVp, 2 mm Al), the MTF went to 0.2 at 11.2 mm(-1) and to 0.1 at 13.3 mm(-1). The DQE reached a maximum value of 54% at 1.35 mm(-1) for the RQA-M2 technique at 1.6 microC/kg and achieved a peak value of 64% at 1.75 mm(-1) for the tungsten technique (MW2) at 1.9 microC/kg. Nevertheless, the DQE showed strong exposure and frequency dependencies. The results indicated that the detector offered high MTFs and DQEs, but structured noise effects may require improved calibration before clinical implementation.

Authors
Saunders, RS; Samei, E; Jesneck, JL; Lo, JY
MLA Citation
Saunders, RS, Samei, E, Jesneck, JL, and Lo, JY. "Physical characterization of a prototype selenium-based full field digital mammography detector." Med Phys 32.2 (February 2005): 588-599.
PMID
15789606
Source
pubmed
Published In
Medical physics
Volume
32
Issue
2
Publish Date
2005
Start Page
588
End Page
599
DOI
10.1118/1.1855033

Issues in assessing multi-institutional performance of BI-RADS-based CAD systems

The purpose of this study was to investigate factors that impact the generalization of breast cancer computer-aided diagnosis (CAD) systems that utilize the Breast Imaging Reporting and Data System (BI-RADS™). Data sets from four institutions were analyzed: Duke University Medical Center, University of Pennsylvania Medical Center, Massachusetts General Hospital, and Wake Forest University. The latter two data sets are subsets of the Digital Database for Screening Mammography. Each data set consisted of descriptions of mammographic lesions according to the BI-RADS lexicon, patient age, and pathology status (benign/malignant). Models were developed to predict pathology status from the BI-RADS descriptors and the patient age. Comparisons between the models built on data from the different institutions were made in terms of empirical (non-parametric) receiver operating characteristic (ROC) curves. Results suggest that BI-RADS-based CAD systems focused on specific classes of lesions may be more generally applicable than models that cover several lesion types. However, better generalization was seen in terms of the area under the ROC curve than in the partial area index (>90% sensitivity). Previous studies have illustrated the challenges in translating a BI-RADS-based CAD system from one institution to another. This study provides new insights into possible approaches to improve the generalization of BI-RADS-based CAD systems.

Authors
Markey, MK; Lo, JY
MLA Citation
Markey, MK, and Lo, JY. "Issues in assessing multi-institutional performance of BI-RADS-based CAD systems." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 5747.II (2005): 858-865.
Source
scival
Published In
Proceedings of SPIE
Volume
5747
Issue
II
Publish Date
2005
Start Page
858
End Page
865
DOI
10.1117/12.594706

Digital breast tomosynthesis using an amorphous selenium flat panel detector

A prototype breast tomosynthesis system* has been developed, allowing a total angular view of ±25°. The detector used in this system is an amorphous selenium direct-conversion digital flat-panel detector suitable for digital tomosynthesis. The system is equipped with various readout sequences to allow the investigation of different tomosynthetic data acquisition modes. In this paper, we will present basic physical properties - such as MTF, NPS, and DQE - measured for the full resolution mode and a binned readout mode of the detector. From the measured projections, slices are reconstructed employing a special version of filtered backprojection algorithm. In a phantom study, we compare binned and full resolution acquisition modes with respect to image quality. Under the condition of same dose, we investigate the impact of the number of views on artifacts. Finally, we show tomosynthesis images reconstructed from first clinical data.

Authors
Bissonnette, M; Hansroul, M; Masson, E; Savard, S; Cadieux, S; Warmoes, P; Gravel, D; Agopyan, J; Polischuk, B; Haerer, W; Mertelmeier, T; Lo, JY; Chen, Y; III, JTD; Jesneck, JL; Singh, S
MLA Citation
Bissonnette, M, Hansroul, M, Masson, E, Savard, S, Cadieux, S, Warmoes, P, Gravel, D, Agopyan, J, Polischuk, B, Haerer, W, Mertelmeier, T, Lo, JY, Chen, Y, III, JTD, Jesneck, JL, and Singh, S. "Digital breast tomosynthesis using an amorphous selenium flat panel detector." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 5745.I (2005): 529-540.
Source
scival
Published In
Proceedings of SPIE
Volume
5745
Issue
I
Publish Date
2005
Start Page
529
End Page
540
DOI
10.1117/12.601622

Detector evaluation of a prototype amorphous selenium-based full field digital mammography system

This study evaluated the physical performance of a selenium-based direct full-field digital mammography prototype detector (Siemens Mammomat Novation DR), including the pixel value vs. exposure linearity, the modulation transfer function (MTF), the normalized noise power spectrum (NNPS), and the detective quantum efficiency (DQE). The current detector is the same model which received an approvable letter from FDA for release to the US market. The results of the current prototype are compared to those of an earlier prototype. Two IEC standard beam qualities (RQA-M2: Mo/Mo, 28 kVp, 2 mm Al; RQA-M4: Mo/Mo, 35 kVp, 2 mm Al) and two additional beam qualities (MW2: W/Rh, 28 kVp, 2 mm Al; MW4: W/Rh, 35 kVp, 2 mm Al) were investigated. To calculate the modulation transfer function (MTF), a 0.1 mm Pt-Ir edge was imaged at each beam quality. Detector pixel values responded linearly against exposure values (R 2 0.999). As before, above 6 cycles/mm Mo/Mo MTF was slightly higher along the chest-nipple axis compared to the left-right axis. MTF was comparable to the previously reported prototype, with slightly reduced resolution. The DQE peaks ranged from 0.71 for 3.31 μC/kg (12.83 mR) to 0.4 for 0.48 μC/kg (1.86 mR) at 1.75 cycles/mm for Mo/Mo at 28 kVp. The DQE range for W/Rh at 28 kVP was 0.81 at 2.03 μC/kg (7.87 mR) to 0.50 at 0.50 μC/kg (1.94 mR) at 1 cycle/mm. NNPS tended to increase with greater exposures, while all exposures had a significant low-frequency component. Bloom and detector edge artifacts observed previously were no longer present in this prototype. The new detector shows marked noise improvement, with slightly reduced resolution. There remain artifacts due to imperfect gain calibration, but at a reduced magnitude compared to a prototype detector.

Authors
Jesneck, JL; Saunders, RS; Samei, E; Xia, JQ; Lo, JY
MLA Citation
Jesneck, JL, Saunders, RS, Samei, E, Xia, JQ, and Lo, JY. "Detector evaluation of a prototype amorphous selenium-based full field digital mammography system." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 5745.I (2005): 478-485.
Source
scival
Published In
Proceedings of SPIE
Volume
5745
Issue
I
Publish Date
2005
Start Page
478
End Page
485
DOI
10.1117/12.596087

Characterization of scatter radiation of a breast phantom on siemens prototype FFDM with and without an anti-scatter grid

In this study, the beam stop technique was applied to obtain the scatter fraction values for an anthropomorphic breast phantom on a flat-panel full field mammography system. The phantom was equivalent to a compressed breast of 5 cm thickness with 50% glandular tissue content. The images were acquired at 28kVp with Mo/Mo target/filter combination and multiple mAs values with or without an anti-scatter grid. The one-dimensional and two-dimensional scatter fraction profiles of the breast phantom without a grid were plotted. The effect of mAs on scatter fraction calculation was investigated. It was found that in order to get a reliable measurement of scatter fraction, the mAs had to be above about 40mAs without a grid and 160mAs with a grid. In addition, the SNR values at 28kVp and 80mAs (AEC level for the phantom using the same imaging technique on a screen/film system) with and without a grid were compared with each other. The SNR with a grid was slightly smaller than that without a grid. The SNR improvement factor (KSNR) defined as the ratio of SNR without the grid to SNR with the grid was 0.976. The grid had the primary and scatter transmissions of 73% and 13% respectively.

Authors
Xia, JQ; Lo, JY; Jr, CEF
MLA Citation
Xia, JQ, Lo, JY, and Jr, CEF. "Characterization of scatter radiation of a breast phantom on siemens prototype FFDM with and without an anti-scatter grid." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 5745.II (2005): 1096-1102.
Source
scival
Published In
Proceedings of SPIE
Volume
5745
Issue
II
Publish Date
2005
Start Page
1096
End Page
1102
DOI
10.1117/12.595960

Impulse response analysis for several digital tomosynthesis mammography reconstruction algorithms

Digital tomosynthesis mammography algorithms allow reconstructions of arbitrary planes in the breast from limited-angle series of projection images as the x-ray source moves along an arc above the breast. Though several tomosynthesis algorithms have been proposed, no complete comparison of the methods has previously been conducted. This paper presents an analysis of impulse response for four different tomosynthesis mammography reconstruction algorithms. Simulated impulses at different 3-D locations were simulated to investigate the sharpness of reconstructed in-plane structures and to see how effective each algorithm is at removing out-of-plane blur. Datasets with 41, 21 and 11 projection images of the impulse were generated with a total angular movement of +/-10 degrees of the simulated x-ray point source. Four algorithms, including shift-and-add method, Niklason algorithm, filtered back projection (FBP), and matrix inversion tomosynthesis (MITS) are investigated. Compared with shift-and-add algorithm and Niklason method, MITS and FBP performed better for in-plane response and out-of-plane blur removal. MITS showed better out-of-plane blur removal in general. MITS and FBP performed better when projection numbers increase.

Authors
Chen, Y; Lo, JY; III, JTD
MLA Citation
Chen, Y, Lo, JY, and III, JTD. "Impulse response analysis for several digital tomosynthesis mammography reconstruction algorithms." Progress in Biomedical Optics and Imaging - Proceedings of SPIE 5745.I (2005): 541-549.
Source
scival
Published In
Proceedings of SPIE
Volume
5745
Issue
I
Publish Date
2005
Start Page
541
End Page
549
DOI
10.1117/12.595684

A framework for optimising the radiographic technique in digital X-ray imaging.

The transition to digital radiology has provided new opportunities for improved image quality, made possible by the superior detective quantum efficiency and post-processing capabilities of new imaging systems, and advanced imaging applications, made possible by rapid digital image acquisition. However, this transition has taken place largely without optimising the radiographic technique used to acquire the images. This paper proposes a framework for optimising the acquisition of digital X-ray images. The proposed approach is based on the signal and noise characteristics of the digital images and the applied exposure. Signal is defined, based on the clinical task involved in an imaging application, as the difference between the detector signal with and without a target present against a representative background. Noise is determined from the noise properties of uniformly acquired images of the background, taking into consideration the absorption properties of the detector. Incident exposure is estimated or otherwise measured free in air, and converted to dose. The main figure of merit (FOM) for optimisation is defined as the signal-difference-to-noise ratio (SdNR) squared per unit exposure or (more preferably) dose. This paper highlights three specific technique optimisation studies that used this approach to optimise the radiographic technique for digital chest and breast applications. In the first study, which was focused on chest radiography with a CsI flat-panel detector, a range of kV(p) (50-150) and filtration (Z = 13-82) were examined in terms of their associated FOM as well as soft tissue to bone contrast, a factor of importance in digital chest radiography. The results indicated that additive Cu filtration can improve image quality. A second study in digital mammography using a selenium direct flat-panel detector indicated improved SdNR per unit exposure with the use of a tungsten target and a rhodium filter than conventional molybdenum target/molybdenum filter techniques. Finally, a third study focusing on cone-beam computed tomography of the breast using a CsI flat-panel detector indicated that high Z filtration of a tungsten target X-ray beam can notably improve the signal and noise characteristics of the image. The general findings highlight the fact that the techniques that are conventionally assumed to be optimum may need to be revisited for digital radiography.

Authors
Samei, E; Dobbins, JT; Lo, JY; Tornai, MP
MLA Citation
Samei, E, Dobbins, JT, Lo, JY, and Tornai, MP. "A framework for optimising the radiographic technique in digital X-ray imaging." Radiat Prot Dosimetry 114.1-3 (2005): 220-229.
PMID
15933112
Source
pubmed
Published In
Radiation Protection Dosimetry
Volume
114
Issue
1-3
Publish Date
2005
Start Page
220
End Page
229
DOI
10.1093/rpd/nch562

Computer-aided detection in screening mammography: variability in cues.

PURPOSE: To evaluate the variability of true-positive and false-positive cues by using a commercially available computer-aided detection (CAD) system for analysis of 50 malignancies in a screening population. MATERIALS AND METHODS: Fifty breast cancers detected at screening were analyzed by using a commercially available CAD system. Mean patient age was 62.2 years. Each set of mammograms (craniocaudal and mediolateral oblique views) was digitized and analyzed by the CAD system 10 times. One radiologist compared CAD output with the location of the malignancy at mammography and determined whether each lesion was marked accurately in one mammographic view, both views, or neither. Sensitivity and reproducibility of the CAD system were determined for both case- and image-based analysis. RESULTS: Overall sensitivity of the CAD system when at least one of the two mammographic views was marked correctly (case-base sensitivity) was 82.4%. Sensitivity when each mammographic view was considered separately (image-based sensitivity) was 61.1%. For case-based analysis, variability in true-positive CAD cues was demonstrated for 14 of 50 (28%) cases. For image-based analysis, inconsistency in CAD output was observed in 33 of 100 (33%) mammographic views that contained malignancies detected at screening. However, the CAD system consistently detected 40-43 of the 50 breast cancers in each of the 10 CAD runs. Variability for false-positive marks was significantly greater than that for true-positive marks. CONCLUSION: Inconsistency was demonstrated for CAD analysis of breast cancers detected at screening. However, the CAD system was reasonably consistent in the overall number of cancers identified from run to run. Greater variability of the CAD system was also demonstrated for false-positive marks, as compared with true-positive marks.

Authors
Baker, JA; Lo, JY; Delong, DM; Floyd, CE
MLA Citation
Baker, JA, Lo, JY, Delong, DM, and Floyd, CE. "Computer-aided detection in screening mammography: variability in cues." Radiology 233.2 (November 2004): 411-417.
PMID
15358850
Source
pubmed
Published In
Radiology
Volume
233
Issue
2
Publish Date
2004
Start Page
411
End Page
417
DOI
10.1148/radiol.2332031200

Fundamental imaging characteristics of a slot-scan digital chest radiographic system.

Our purpose in this study was to evaluate the fundamental image quality characteristics of a new slot-scan digital chest radiography system (ThoraScan, Delft Imaging Systems/Nucletron, Veenendaal, The Netherlands). The linearity of the system was measured over a wide exposure range at 90, 117, and 140 kVp with added Al filtration. System uniformity and reproducibility were established with an analysis of images from repeated exposures. The modulation transfer function (MTF) was evaluated using an established edge method. The noise power spectrum (NPS) and the detective quantum efficiency (DQE) of the system were evaluated at the three kilo-voltages over a range of exposures. Scatter fraction (SF) measurements were made using a posterior beam stop method and a geometrical chest phantom. The system demonstrated excellent linearity, but some structured nonuniformities. The 0.1 MTF values occurred between 3.3-3.5 mm(-1). The DQE(0.15) and DQE(2.5) were 0.21 and 0.07 at 90 kVp, 0.18 and 0.05 at 117 kVp, and 0.16 and 0.03 at 140 kVp, respectively. The system exhibited remarkably lower SFs compared to conventional full-field systems with anti-scatter grid, measuring 0.13 in the lungs and 0.43 in the mediastinum. The findings indicated that the slot-scan design provides marked scatter reduction leading to high effective DQE (DQEeff) of the system and reduced patient dose required to achieve high image quality.

Authors
Samei, E; Saunders, RS; Lo, JY; Dobbins, JT; Jesneck, JL; Floyd, CE; Ravin, CE
MLA Citation
Samei, E, Saunders, RS, Lo, JY, Dobbins, JT, Jesneck, JL, Floyd, CE, and Ravin, CE. "Fundamental imaging characteristics of a slot-scan digital chest radiographic system." Med Phys 31.9 (September 2004): 2687-2698.
PMID
15487752
Source
pubmed
Published In
Medical physics
Volume
31
Issue
9
Publish Date
2004
Start Page
2687
End Page
2698
DOI
10.1118/1.1783531

New results in computer aided diagnosis (CAD) of breast cancer using a recently developed SVM/GRNN oracle hybrid

Breast cancer is second only to lung cancer as a tumor-related cause of death in women. Currently, the method of choice for the early detection of breast cancer is mammography. While sensitive to the detection of non palpable breast lesions, its positive predictive value (PPV) is low, resulting in biopsies that are only 15%-34% likely to reveal malignancy. This paper explores the use of a recently designed Support Vector Machine (SVM)/Generalized Regression Neural Network (GRNN) Oracle hybrid to classify breast lesions and evaluate the software's performance as an interpretive aid to radiologists. The main objective of the research was to perform an independent analysis, using a new, integrated film screen mammogram data base of approximately 2500 cases from five separate institutions, to verify results obtained previously 14. This study demonstrated the following: The DE crossover constant has little, if any, effect on measures of performance (MOP). A specificity of approximately 5.6% is achieved at 100% sensitivity, which increases to approximately 36% at 95% sensitivity. PPV increases from 51 % to 56% as sensitivity is decreased from 100 to 95%, respectively.

Authors
Jr, WHL; Wong, L; McKee, D; Masters, T; Anderson, F; Raturi, A; Lo, JY
MLA Citation
Jr, WHL, Wong, L, McKee, D, Masters, T, Anderson, F, Raturi, A, and Lo, JY. "New results in computer aided diagnosis (CAD) of breast cancer using a recently developed SVM/GRNN oracle hybrid." Proceedings of SPIE - The International Society for Optical Engineering 5370 II (2004): 777-784.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5370 II
Publish Date
2004
Start Page
777
End Page
784
DOI
10.1117/12.533142

Breast cancer classification improvements using a new kernel function with evolutionary-programming-configured support vector machines

Mammography is an effective tool for the early detection of breast cancer; however, most women referred for biopsy based on mammographic findings do not, in fact, have cancer. This study is part of an ongoing effort to reduce the number of benign cases referred for biopsy by developing tools to aid physicians in classifying suspicious lesions. Specifically, this study examines the use of an Evolutionary Programming (EP)-derived Support Vector Machine (SVM) with a modified radial basis function (RBF) kernel, and compares this with results using a normal Gaussian radial basis function kernel. Results demonstrate that the modified kernel can provide moderate performance improvements; however, due to its ability to create a more complex decision surface, this kernel can easily begin to memorize the training data resulting in a loss of generalization ability. Nonetheless, these methods could reduce the number of benign cases referred for biopsy by over half, while missing less than 5% of malignancies. Future work will focus on methods to improve the EP process to preserve SVMs which generalize well.

Authors
Jr, WHL; McKee, DW; Anderson, FR; Lo, JY
MLA Citation
Jr, WHL, McKee, DW, Anderson, FR, and Lo, JY. "Breast cancer classification improvements using a new kernel function with evolutionary-programming-configured support vector machines." Proceedings of SPIE - The International Society for Optical Engineering 5370 II (2004): 880-887.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5370 II
Publish Date
2004
Start Page
880
End Page
887
DOI
10.1117/12.535864

Bayesian networks of BI-RADS™ descriptors for breast lesion Classification

We investigated Bayesian network structure learning and probability estimation from mammographic feature data in order to classify breast lesions into different pathological categories. We compared the learned networks to naïve Bayes classifiers, which are similar to the expert systems previously investigated for breast lesion classification. The learned network structures reflect the difference in the classification of biopsy outcome and the invasiveness of malignant lesions for breast masses and microcalcifications. The difference between masses and microcalcifications should be taken into consideration when interpreting systems for automatic pathological classification of breast lesions. The difference may also affect use of these systems for tasks such as estimating the sampling error of biopsy.

Authors
Fischer, EA; Lo, JY; Markey, MK
MLA Citation
Fischer, EA, Lo, JY, and Markey, MK. "Bayesian networks of BI-RADS™ descriptors for breast lesion Classification." Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings 26 IV (2004): 3031-3034.
Source
scival
Published In
Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings
Volume
26 IV
Publish Date
2004
Start Page
3031
End Page
3034

Computer-aided classification of breast masses using mammogram, ultrasound, and clinical inputs

Authors
Hong, AS; Baker, JA; Lo, JY; Nicholas, JL; Soo, MS
MLA Citation
Hong, AS, Baker, JA, Lo, JY, Nicholas, JL, and Soo, MS. "Computer-aided classification of breast masses using mammogram, ultrasound, and clinical inputs." 2004.
Source
wos-lite
Published In
AJR. American journal of roentgenology
Volume
182
Issue
4
Publish Date
2004
Start Page
33
End Page
33

Computer-aided detection (CAD) in screening mammography: sensitivity of commercial CAD systems for detecting architectural distortion.

OBJECTIVE: Computer-aided detection (CAD) algorithms have successfully revealed breast masses and microcalcifications on screening mammography. The purpose of our study was to evaluate the sensitivity of commercially available CAD systems for revealing architectural distortion, the third most common appearance of breast cancer. MATERIALS AND METHODS: Two commercially available CAD systems were used to evaluate screening mammograms obtained in 43 patients with 45 mammographically detected regions of architectural distortion. For each CAD system, we determined the sensitivity for revealing architectural distortion on at least one image of the two-view mammographic examination (case sensitivity) and for each individual mammogram (image sensitivity). Surgical biopsy results were available for each case of architectural distortion. RESULTS: Architectural distortion was deemed present and actionable by a panel of expert breast imagers in 80 views of the 45 cases. One CAD system detected distortion in 22 of 45 cases of distortion (case sensitivity, 49%) and in 30 of 80 mammograms (image sensitivity, 38%); it displayed 0.7 false-positive marks per image. Another CAD system identified distortion in 15 of 45 cases (case sensitivity, 33%) and 17 of 80 mammograms (image sensitivity, 21%); it displayed 1.27 false-positive marks per image. Sensitivity for malignancy-caused distortion was similar to or lower than sensitivity for all causes of distortion. CONCLUSION: Fewer than one half of the cases of architectural distortion were detected by the two most widely available CAD systems used for interpretations of screening mammograms. Considerable improvement in the sensitivity of CAD systems is needed for detecting this type of lesion. Practicing breast imagers who use CAD systems should remain vigilant for architectural distortion.

Authors
Baker, JA; Rosen, EL; Lo, JY; Gimenez, EI; Walsh, R; Soo, MS
MLA Citation
Baker, JA, Rosen, EL, Lo, JY, Gimenez, EI, Walsh, R, and Soo, MS. "Computer-aided detection (CAD) in screening mammography: sensitivity of commercial CAD systems for detecting architectural distortion." AJR Am J Roentgenol 181.4 (October 2003): 1083-1088.
PMID
14500236
Source
pubmed
Published In
AJR. American journal of roentgenology
Volume
181
Issue
4
Publish Date
2003
Start Page
1083
End Page
1088
DOI
10.2214/ajr.181.4.1811083

Application of likelihood ratio to classification of mammographic masses; performance comparison to case-based reasoning.

The likelihood ratio (LR) is an optimal approach for deciding which of two alternate hypotheses best describes a given situation. We adopted this formalism for predicting whether biopsy results of mammographic masses will be benign or malignant, aiming to reduce the number of biopsies performed on benign lesions. We compared the performance of this LR-based algorithm (LRb) to a case-based reasoning (CBR) classifier, which provides a solution to a new problem using past similiar cases. Each classifier used mammographers' BI-RADS descriptions of mammographic masses as input. The database consisted of 646 biopsy-proven mammography cases. Performance was evaluated using Receiver Operating Characteristic (ROC) analysis, Round Robin sampling, and bootstrap. The ROC areas (AUC) for the LRb and CBR were 0.91+/- 0.01 and 0.92 +/- 0.01, respectively. The partial ROC area index (0.90AUC) was the same for both classifiers, 0.59 +/- 0.05. At a sensitivity of 98%, the CBR would spare 204 (49%) of benign lesions from biopsy; the LRb would spare 209 (51%) benign lesions. The performance of the two classifiers was very similar, with no statistical differences in AUC or 0.90AUC. Although the CBR and LRb originate from different fields of study, their implementations differ only in the estimation of the probability density functions (PDFs) of the feature distributions. The CBR performs this estimation implicitly, while using various similarity metrics. On the other hand, the estimation of the PDFs is specified explicitly in the LRb implementation. This difference in the estimation of the PDFs results in the very small difference in performance, and at 98% sensitivity, both classifiers would spare about half of the benign mammographic masses from biopsy. The CBR and LRb are equivalent methods in implementation and performance.

Authors
Bilska-Wolak, AO; Floyd, CE; Nolte, LW; Lo, JY
MLA Citation
Bilska-Wolak, AO, Floyd, CE, Nolte, LW, and Lo, JY. "Application of likelihood ratio to classification of mammographic masses; performance comparison to case-based reasoning." Med Phys 30.5 (May 2003): 949-958.
PMID
12773004
Source
pubmed
Published In
Medical physics
Volume
30
Issue
5
Publish Date
2003
Start Page
949
End Page
958
DOI
10.1118/1.1565339

Self-organizing map for cluster analysis of a breast cancer database.

The purpose of this study was to identify and characterize clusters in a heterogeneous breast cancer computer-aided diagnosis database. Identification of subgroups within the database could help elucidate clinical trends and facilitate future model building. A self-organizing map (SOM) was used to identify clusters in a large (2258 cases), heterogeneous computer-aided diagnosis database based on mammographic findings (BI-RADS) and patient age. The resulting clusters were then characterized by their prototypes determined using a constraint satisfaction neural network (CSNN). The clusters showed logical separation of clinical subtypes such as architectural distortions, masses, and calcifications. Moreover, the broad categories of masses and calcifications were stratified into several clusters (seven for masses and three for calcifications). The percent of the cases that were malignant was notably different among the clusters (ranging from 6 to 83%). A feed-forward back-propagation artificial neural network (BP-ANN) was used to identify likely benign lesions that may be candidates for follow up rather than biopsy. The performance of the BP-ANN varied considerably across the clusters identified by the SOM. In particular, a cluster (#6) of mass cases (6% malignant) was identified that accounted for 79% of the recommendations for follow up that would have been made by the BP-ANN. A classification rule based on the profile of cluster #6 performed comparably to the BP-ANN, providing approximately 25% specificity at 98% sensitivity. This performance was demonstrated to generalize to a large (2177) set of cases held-out for model validation.

Authors
Markey, MK; Lo, JY; Tourassi, GD; Floyd, CE
MLA Citation
Markey, MK, Lo, JY, Tourassi, GD, and Floyd, CE. "Self-organizing map for cluster analysis of a breast cancer database." Artif Intell Med 27.2 (February 2003): 113-127.
PMID
12636975
Source
pubmed
Published In
Artificial Intelligence in Medicine
Volume
27
Issue
2
Publish Date
2003
Start Page
113
End Page
127

Improving the predictive value of mammography using a specialized evolutionary programming hybrid and fitness functions

Mammography is an effective tool for the early detection of breast cancer; however, most women referred for biopsy based on mammographic findings do not, have cancer. This study is part of an ongoing effort to reduce the number of benign cases referred for biopsy by developing tools to aid physicians in classifying suspicious lesions. Specifically, this study examines the use of an Evolutionary Programming (EP)/Adaptive Boosting (AB) hybrid, specifically modified to focus on improving the performance of computer-assisted diagnostic (CAD) tools at high specificity levels (missing few or no cancers). An EP/AB hybrid developed by the authors and used in previous studies was modified with two new fitness functions: 1) a function which favored networks with high PPV values at thresholds corresponding to high sensitivities, and 2) a function which favored networks with the highest partial ROC Az (normalized area above 90% sensitivity). The modified hybrid with specialized fitness functions was evaluated using k-fold cross-validation against two real-word mammogram data sets. Results indicate that the number of benign cases referred for biopsy might be reduced by over a third, while missing no cancers. If sensitivity is allowed to decrease to 97% (missing 3% of the cancers), the number of spared biopsies could be raised to over half.

Authors
Jr, WHL; McKee, DW; Lo, JY; Anderson, FR
MLA Citation
Jr, WHL, McKee, DW, Lo, JY, and Anderson, FR. "Improving the predictive value of mammography using a specialized evolutionary programming hybrid and fitness functions." Proceedings of SPIE - The International Society for Optical Engineering 5032 II (2003): 898-907.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5032 II
Publish Date
2003
Start Page
898
End Page
907
DOI
10.1117/12.483550

Application of support vector machines to breast cancer screening using mammogram and clinical history data

The objectives of this paper are to discuss: (1) the development and testing of a new Evolutionary Programming (EP) method to optimally configure Support Vector Machine (SVM) parameters for facilitating the diagnosis of breast cancer; (2) evaluation of EP derived learning machines when the number of BI-RADS™ and clinical history discriminators are reduced from 16 to 7; (3) establishing system performance for several SVM kernels in addition to the EP/Adaptive Boosting (EP/AB) hybrid using the Digital Database for Screening Mammography, University of South Florida (DDSM USF) and Duke data sets; and (4) obtaining a preliminary evaluation of the measurement of SVM learning machine inter-institutional generalization capability using BI-RADS™ data. Measuring performance of the SVM designs and EP/AB hybrid against these objectives will provide quantative evidence that the software packages described can generalize to larger patient data sets from different institutions. Most iterative methods currently in use to optimize learning machine parameters are time consuming processes, which sometimes yield sub-optimal values resulting in performance degradation. SVMs are new machine Intelligence paradigms, which use the Structural Risk Minimization (SRM) concept to develop learning machines. These learning machines can always be trained to provide global minima, given that the machine parameters are optimally computed. In addition, several system performance studies are described which include EP derived SVM performance as a function of: (a) population and generation size as well as a method for generating initial populations and (b) iteratively derived versus EP derived learning machine parameters. Finally, the authors describe a set of experiments providing preliminary evidence that both the EP/AB hybrid and SVM Computer Aided Diagnostic C++ software packages will work across a large population of patients, based on a data set of approximately 2,500 samples from five different institutions.

Authors
Jr, WHL; McKee, D; Velazquez, R; Wong, L; Lo, JY; Anderson, F
MLA Citation
Jr, WHL, McKee, D, Velazquez, R, Wong, L, Lo, JY, and Anderson, F. "Application of support vector machines to breast cancer screening using mammogram and clinical history data." Proceedings of SPIE - The International Society for Optical Engineering 5032 I (2003): 546-556.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5032 I
Publish Date
2003
Start Page
546
End Page
556
DOI
10.1117/12.480235

Prediction of breast biopsy outcome using a likelihood ratio classifier and biopsy cases from two medical centers

Potential malignancy of a mammographie lesion can be assessed using the mathematically optimal likelihood ratio (LR) from signal detection theory. We developed a LR classifier for prediction of breast biopsy outcome of mammographie masses from BI-RADS findings. We used cases from Duke University Medical Center (645 total, 232 malignant) and University of Pennsylvania (496, 200). The LR was trained and tested alternatively on both subsets. Leave-one-out sampling was used when training and testing was performed on the same data set. When tested on the Duke set, the LR achieved a Received Operating Characteristic (ROC) area of 0.91 ± 0.01, regardless of whether Duke or Pennsylvania set was used for training. The LR achieved a ROC area of 0.85 ± 0.02 for the Pennsylvania set, again regardless of which set was used for training. When using actual case data for training, the LR's procedure is equivalent to case-based reasoning, and can explain the classifier's decisions in terms of similarity to other cases. These preliminary results suggest that the LR is a robust classifier for prediction of biopsy outcome using biopsy cases from different medical centers.

Authors
Bilska-Wolak, AO; Jr, CEF; Lo, JY
MLA Citation
Bilska-Wolak, AO, Jr, CEF, and Lo, JY. "Prediction of breast biopsy outcome using a likelihood ratio classifier and biopsy cases from two medical centers." Proceedings of SPIE - The International Society for Optical Engineering 5032 III (2003): 1386-1391.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5032 III
Publish Date
2003
Start Page
1386
End Page
1391
DOI
10.1117/12.481349

Computer-aided classification of breast microcalcification clusters: Merging of features from image processing and radiologists

We developed an ensemble classifier for the task of computer-aided diagnosis of breast microcalcification clusters, which are very challenging to characterize for radiologists and computer models alike. The purpose of this study is to help radiologists identify whether suspicious calcification clusters are benign vs. malignant, such that they may potentially recommend fewer unnecessary biopsies for actually benign lesions. The data consists of mammographic features extracted by automated image processing algorithms as well as manually interpreted by radiologists according to a standardized lexicon. We used 292 cases from a publicly available mammography database. From each cases, we extracted 22 image processing features pertaining to lesion morphology, 5 radiologist features also pertaining to morphology, and the patient age. Linear discriminant analysis (LDA) models were designed using each of the three data types. Each local model performed poorly; the best was one based upon image processing features which yielded ROC area index Az of 0.59 ± 0.03 and partial Az above 90% sensitivity of 0.08 ± 0.03. We then developed ensemble models using different combinations of those data types, and these models all improved performance compared to the local models. The final ensemble model was based upon 5 features selected by stepwise LDA from all 28 available features. This ensemble performed with A z of 0.69 ± 0.03 and partial Az of 0.21 ± 0.04, which was statistically significantly better than the model based on the image processing features alone (p<0.001 and p=0.01 for full and partial Az respectively). This demonstrated the value of the radiologist-extracted features as a source of information for this task. It also suggested there is potential for improved performance using this ensemble classifier approach to combine different sources of currently available data.

Authors
Lo, JY; Gavrielides, M; Markey, MK; Jesneck, JL
MLA Citation
Lo, JY, Gavrielides, M, Markey, MK, and Jesneck, JL. "Computer-aided classification of breast microcalcification clusters: Merging of features from image processing and radiologists." Proceedings of SPIE - The International Society for Optical Engineering 5032 II (2003): 882-889.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5032 II
Publish Date
2003
Start Page
882
End Page
889
DOI
10.1117/12.480869

Validation of a constraint satisfaction neural network for breast cancer diagnosis: New results from 1,030 cases

Previously, we presented a Constraint Satisfaction Neural Network (CSNN) to predict the outcome of breast biopsy using mammographic and clinical findings. Based on 500 cases, the study showed that CSNN was able to operate not only as a predictive but also as a knowledge discovery tool. The purpose of this study is to validate the CSNN on a database of additional 1,030 cases. An auto-associative backpropagation scheme was used to determine the CSNN constraints based on the initial 500 patients. Subsequently, the CSNN was applied to 1,030 new patients (358 patients with malignant and 672 with benign lesions) to predict breast lesion malignancy. For every test case, the CSNN reconstructed the diagnosis node given the network constraints and the external inputs to the network. The activation level achieved by the diagnosis node was used as the decision variable for ROC analysis. Overall, the CSNN continued to perform well over this large dataset with ROC area of Az = 0.81 ± 0.02. However, the diagnostic performance of the network was inferior in cases with missing clinical findings (Az = 0.80 ± 0.02) compared to those with complete findings (Az = 0.84 ± 0.03). The study also demonstrated the ability of the CSNN to effectively impute missing findings while performing as a predictive tool.

Authors
Tourassi, GD; Lo, JY; Markey, MK
MLA Citation
Tourassi, GD, Lo, JY, and Markey, MK. "Validation of a constraint satisfaction neural network for breast cancer diagnosis: New results from 1,030 cases." Proceedings of SPIE - The International Society for Optical Engineering 5032 I (2003): 207-214.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
5032 I
Publish Date
2003
Start Page
207
End Page
214
DOI
10.1117/12.481111

Differences between computer-aided diagnosis of breast masses and that of calcifications.

PURPOSE: To compare the performance of a computer-aided diagnosis (CAD) system for diagnosis of previously detected lesions, based on radiologist-extracted findings on masses and calcifications. MATERIALS AND METHODS: A feed-forward, back-propagation artificial neural network (BP-ANN) was trained in a round-robin (leave-one-out) manner to predict biopsy outcome from mammographic findings (according to the Breast Imaging Reporting and Data System) and patient age. The BP-ANN was trained by using a large (>1,000 cases) heterogeneous data set containing masses and microcalcifications. The performances of the BP-ANN on masses and microcalcifications were compared with use of receiver operating characteristic analysis and a z test for uncorrelated samples. RESULTS: The BP-ANN performed significantly better on masses than microcalcifications in terms of both the area under the receiver operating characteristic curve and the partial receiver operating characteristic area index. A similar difference in performance was observed with a second model (linear discriminant analysis) and also with a second data set from a similar institution. CONCLUSION: Masses and calcifications should be considered separately when evaluating CAD systems for breast cancer diagnosis.

Authors
Markey, MK; Lo, JY; Floyd, CE
MLA Citation
Markey, MK, Lo, JY, and Floyd, CE. "Differences between computer-aided diagnosis of breast masses and that of calcifications." Radiology 223.2 (May 2002): 489-493.
PMID
11997558
Source
pubmed
Published In
Radiology
Volume
223
Issue
2
Publish Date
2002
Start Page
489
End Page
493
DOI
10.1148/radiol.2232011257

Parameter optimization of a computer-aided diagnosis scheme for the segmentation of microcalcification clusters in mammograms.

Our purpose in this study is to develop a parameter optimization technique for the segmentation of suspicious microcalcification clusters in digitized mammograms. In previous work, a computer-aided diagnosis (CAD) scheme was developed that used local histogram analysis of overlapping subimages and a fuzzy rule-based classifier to segment individual microcalcifications, and clustering analysis for reducing the number of false positive clusters. The performance of this previous CAD scheme depended on a large number of parameters such as the intervals used to calculate fuzzy membership values and on the combination of membership values used by each decision rule. These parameters were optimized empirically based on the performance of the algorithm on the training set. In order to overcome the limitations of manual training and rule generation, the segmentation algorithm was modified in order to incorporate automatic parameter optimization. For the segmentation of individual microcalcifications, the new algorithm used a neural network with fuzzy-scaled inputs. The fuzzy-scaled inputs were created by processing the histogram features with a family of membership functions, the parameters of which were automatically extracted from the distribution of the feature values. The neural network was trained to classify feature vectors as either positive or negative. Individual microcalcifications were segmented from positive subimages. After clustering, another neural network was trained to eliminate false positive clusters. A database of 98 images provided training and testing sets to optimize the parameters and evaluate the CAD scheme, respectively. The performance of the algorithm was evaluated with a FROC analysis. At a sensitivity rate of 93.2%, there was an average of 0.8 false positive clusters per image. The results are very comparable with those taken using our previously published rule-based method. However, the new algorithm is more suited to generalize its performance on a larger population, depends on two monotonic outputs making its evaluation much easier and can be trained in an automatic way making practical its application on a large database.

Authors
Gavrielides, MA; Lo, JY; Floyd, CE
MLA Citation
Gavrielides, MA, Lo, JY, and Floyd, CE. "Parameter optimization of a computer-aided diagnosis scheme for the segmentation of microcalcification clusters in mammograms." Med Phys 29.4 (April 2002): 475-483.
PMID
11998828
Source
pubmed
Published In
Medical physics
Volume
29
Issue
4
Publish Date
2002
Start Page
475
End Page
483
DOI
10.1118/1.1460874

Outcome analysis of patients with acute pancreatitis by using an artificial neural network.

RATIONALE AND OBJECTIVES: The authors performed this study to evaluate the ability of an artificial neural network (ANN) that uses radiologic and laboratory data to predict the outcome in patients with acute pancreatitis. MATERIALS AND METHODS: An ANN was constructed with data from 92 patients with acute pancreatitis who underwent computed tomography (CT). Input nodes included clinical, laboratory, and CT data. The ANN was trained and tested by using a round-robin technique, and the performance of the ANN was compared with that of linear discriminant analysis and Ranson and Balthazar grading systems by using receiver operating characteristic analysis. The length of hospital stay was used as an outcome measure. RESULTS: Hospital stay ranged from 0 to 45 days, with a mean of 8.4 days. The hospital stay was shorter than the mean for 62 patients and longer than the mean for 30. The 23 input features were reduced by using stepwise linear discriminant analysis, and an ANN was developed with the six most statistically significant parameters (blood pressure, extent of inflammation, fluid aspiration, serum creatinine level, serum calcium level, and the presence of concurrent severe illness). With these features, the ANN successfully predicted whether the patient would exceed the mean length of stay (Az = 0.83 +/- 0.05). Although the Az performance of the ANN was statistically significantly better than that of the Ranson (Az = 0.68 +/- 0.06, P < .02) and Balthazar (Az = 0.62 +/- 0.06, P < .003) grades, it was not significantly better than that of linear discriminant analysis (Az = 0.82 +/- 0.05, P = .53). CONCLUSION: An ANN may be useful for predicting outcome in patients with acute pancreatitis.

Authors
Keogan, MT; Lo, JY; Freed, KS; Raptopoulos, V; Blake, S; Kamel, IR; Weisinger, K; Rosen, MP; Nelson, RC
MLA Citation
Keogan, MT, Lo, JY, Freed, KS, Raptopoulos, V, Blake, S, Kamel, IR, Weisinger, K, Rosen, MP, and Nelson, RC. "Outcome analysis of patients with acute pancreatitis by using an artificial neural network." Acad Radiol 9.4 (April 2002): 410-419.
PMID
11942655
Source
pubmed
Published In
Academic Radiology
Volume
9
Issue
4
Publish Date
2002
Start Page
410
End Page
419

Perceptron error surface analysis: a case study in breast cancer diagnosis.

Perceptrons are typically trained to minimize mean square error (MSE). In computer-aided diagnosis (CAD), model performance is usually evaluated according to other more clinically relevant measures. The purpose of this study was to investigate the relationship between MSE and the area (A(z)) under the receiver operating characteristic (ROC) curve and the high-sensitivity partial ROC area ((0.90)A'(z)). A perceptron was used to predict lesion malignancy based on two mammographic findings and patient age. For each performance measure, the error surface in weight space was visualized. Comparison of the surfaces indicated that minimizing MSE tended to maximize A(z), but not (0.90)A'(z).

Authors
Markey, MK; Lo, JY; Vargas-Voracek, R; Tourassi, GD; Floyd, CE
MLA Citation
Markey, MK, Lo, JY, Vargas-Voracek, R, Tourassi, GD, and Floyd, CE. "Perceptron error surface analysis: a case study in breast cancer diagnosis." Comput Biol Med 32.2 (March 2002): 99-109.
PMID
11879823
Source
pubmed
Published In
Computers in Biology and Medicine
Volume
32
Issue
2
Publish Date
2002
Start Page
99
End Page
109

Cross-institutional evaluation of BI-RADS predictive model for mammographic diagnosis of breast cancer.

OBJECTIVE: Given a predictive model for identifying very likely benign breast lesions on the basis of Breast Imaging Reporting and Data System (BI-RADS) mammographic findings, this study evaluated the model's ability to generalize to a patient data set from a different institution. MATERIALS AND METHODS: The artificial neural network model underwent three trials: it was optimized over 500 biopsy-proven lesions from Duke University Medical Center or "Duke," evaluated on 1,000 similar cases from the University of Pennsylvania Health System or "Penn," and reoptimized for Penn. RESULTS: Trial A's Duke-only model yielded 98% sensitivity, 36% specificity, area index (A(z)) of 0.86, and partial A(z) of 0.51. The cross-institutional trial B yielded 96% sensitivity, 28% specificity, A(z) of 0.79, and partial A(z) of 0.28. The decreases were significant for both A(z) (p = 0.017) and partial A(z) (p < 0.001). In trial C, the model reoptimized for the Penn data yielded 96% sensitivity, 35% specificity, A(z) of 0.83, and partial A(z) of 0.32. There were no significant differences compared with trial B for specificity (p = 0.44) or partial A(z) (p = 0.46), suggesting that the Penn data were inherently more difficult to characterize. CONCLUSION: The BI-RADS lexicon facilitated the cross-institutional test of a breast cancer prediction model. The model generalized reasonably well, but there were significant performance decreases. The cross-institutional performance was encouraging because it was not significantly different from that of a reoptimized model using the second data set at high sensitivities. This study indicates the need for further work to collect more data and to improve the robustness of the model.

Authors
Lo, JY; Markey, MK; Baker, JA; Floyd, CE
MLA Citation
Lo, JY, Markey, MK, Baker, JA, and Floyd, CE. "Cross-institutional evaluation of BI-RADS predictive model for mammographic diagnosis of breast cancer." AJR Am J Roentgenol 178.2 (February 2002): 457-463.
PMID
11804918
Source
pubmed
Published In
AJR. American journal of roentgenology
Volume
178
Issue
2
Publish Date
2002
Start Page
457
End Page
463
DOI
10.2214/ajr.178.2.1780457

Performance tradeoff between evolutionary computation (EC)/adaptive boosting (AB) hybrid and support vector machine breast cancer classification paradigms

This paper describes a breast cancer classification performance trade-off analysis using two computational intelligence paradigms. The first, an evolutionary programming (EP)/adaptive boosting (AB) based hybrid, intelligently combines the outputs from an iteratively "called" weak learning algorithm (one which performs at least slightly better than random guessing) in order to "boost" the performance of an EP-derived weak learner. The second paradigm is support vector machines (SVMs). SVMs are new and radically different types of classifiers and learning machines that use a hypothesis space of linear functions in a high dimensional feature space. The most important advantage of a SVM, unlike neural networks, is that SVM training always finds a global minimum. Furthermore, the SVM has inherent ability to solve pattern classification without incorporating any problem-domain knowledge. In this study, the both the EP/AB hybrid and SVM were employed as pattern classifiers, operating on mammography data used for breast cancer detection. The main focus of the study was to construct and seek the best EP/AB hybrid and SVM configurations for optimum specificity and positive predictive value at very high sensitivities. Using a mammogram database of 500 biopsy-proven samples, the best performing SVM, on average, was able to achieve (under statistical 5-fold cross-validation) a specificity of 45.0% and a positive predictive value (PPV) of 50.1% at 100% sensitivity. At 97% sensitivity, a specificity of 55.8% and a PPV of 55.2% were obtained. The best performing EP/AB hybrid obtained slightly lower, but comparable, results. © 2002 IEEE.

Authors
Land, WH; Bryden, M; Lo, JY; McKee, DW; Anderson, FR
MLA Citation
Land, WH, Bryden, M, Lo, JY, McKee, DW, and Anderson, FR. "Performance tradeoff between evolutionary computation (EC)/adaptive boosting (AB) hybrid and support vector machine breast cancer classification paradigms." Proceedings of the 2002 Congress on Evolutionary Computation, CEC 2002 1 (January 1, 2002): 187-192.
Source
scopus
Published In
Proceedings of the 2002 Congress on Evolutionary Computation, CEC 2002
Volume
1
Publish Date
2002
Start Page
187
End Page
192
DOI
10.1109/CEC.2002.1006231

Cluster analysis of BI-RADS™ descriptions of biopsy-proven breast lesions

The purpose of this study was to identify and characterize clusters in a heterogeneous breast cancer computer-aided diagnosis database. Identification of subgroups within the database could help elucidate clinical trends and facilitate future model building. Agglomerative hierarchical clustering and k-means clustering were used to identify clusters in a large, heterogeneous computer-aided diagnosis database based on mammographic findings (BI-RADS™) and patient age. The clusters were examined in terms of their feature distributions. The clusters showed logical separation of distinct clinical subtypes such as architectural distortions, masses, and calcifications. Moreover, the common subtypes of masses and calcifications were stratified into clusters based on age groupings. The percent of the cases that were malignant was notably different among the clusters. Cluster analysis can provide a powerful tool in discerning the subgroups present in a large, heterogeneous computer-aided diagnosis database.

Authors
Markey, MK; Lo, JY; Tourassi, GD; Jr, CEF
MLA Citation
Markey, MK, Lo, JY, Tourassi, GD, and Jr, CEF. "Cluster analysis of BI-RADS™ descriptions of biopsy-proven breast lesions." Proceedings of SPIE - The International Society for Optical Engineering 4684 I (2002): 363-370.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
4684 I
Publish Date
2002
Start Page
363
End Page
370
DOI
10.1117/12.467177

Application of support vector machines to breast cancer screening using mammogram and history data

Support Vector Machines (SVMs) are a new and radically different type of classifiers and learning machines that use a hypothesis space of linear functions in a high dimensional feature space. This relatively new paradigm, based on Statistical Learning theory (SLT) and Structural Risk Minimization (SRM), has many advantages when compared to traditional neural networks, which are based on Empirical Risk Minimization (ERM). Unlike neural networks, SVM training always finds a global minimum. Furthermore, SVMs have inherent ability to solve pattern classification without incorporating any problem-domain knowledge. In this study, the SVM was employed as a pattern classifier, operating on mammography data used for breast cancer detection. The main focus was to formulate the best learning machine configurations for optimum specificity and positive predictive value at very high sensitivities. Using a mammogram database of 500 biopsy-proven samples, the best performing SVM, on average, was able to achieve (under statistical 5-fold cross-validation) a specificity of 45.0% and a positive predictive value (PPV) of 50.1% at 100% sensitivity. At 97% sensitivity, a specificity of 55.8% and a PPV of 55.2% were obtained.

Authors
Jr, WHL; Akanda, A; Lo, JY; Anderson, F; Bryden, M
MLA Citation
Jr, WHL, Akanda, A, Lo, JY, Anderson, F, and Bryden, M. "Application of support vector machines to breast cancer screening using mammogram and history data." Proceedings of SPIE - The International Society for Optical Engineering 4684 I (2002): 636-642.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
4684 I
Publish Date
2002
Start Page
636
End Page
642
DOI
10.1117/12.467206

Improving mammogram screening using a bank of support vector machines (SVMs)

The focus of this study was to build and evaluate a new bank of SVM designs to address the problem of high false positives that currently results from mammogram screening,. The basis of the design is to partition the BIRADS™ varibles into three separate categories based on our understanding of the discriminating information contained in the mammogram BIRADS™ findings. That is, after ascertaining the presence of a suspecious finding on a mammogram that would be recommended for biopsy, the radiologist documents the BIRADS™ lesion descriptor values. This information, along with the clinical history, would be used as imput to this new bank of SVMs an aid to the physican for improving the specificity and positive predictive value PPV of the benign/malignant diagnosis task. Comparing the new SVM mass classifier with the previously configured single SVM that used all data base imputs provided significant classification accuracy improvemens for all performance measures. That is, overall Az improved by 11.6%, specificity and PPV improved by 110.6% and 31.6%, respecitvely, at 100% sensitivity (missing no cancers), while specificity and PPV improved by 54% and 35.9%, respectively, at 95% sensitivity (missing 5% of the cancers).

Authors
Jr, WHL; McKee, DW; Lo, JY; Anderson, F
MLA Citation
Jr, WHL, McKee, DW, Lo, JY, and Anderson, F. "Improving mammogram screening using a bank of support vector machines (SVMs)." Intelligent Engineering Systems Through Artificial Neural Networks 12 (2002): 779-784.
Source
scival
Published In
Intelligent Engineering Systems Through Artificial Neural Networks
Volume
12
Publish Date
2002
Start Page
779
End Page
784

Using evolutionary programming to configure support vector machines for the diagnosis of breast cancer

Support Vector Machines(s) (SVMs) are new machine intelligence paradigms that use the Structural Risk Minimization (SRM) concept to develop learning machines. SVMs can always be trained to provide global minima, given that the leaning machine parameters are optimally computed. The current most prevalent methods to select these parameters are numerical iterative techniques. While useful, these methods frequently have no basis in theory, and cannot guarantee that the resultant parameters will yield optimum learning machine performance. The purpose of this paper is to discuss and demonstrate the application of Evolutionary Programming (EP) concepts to develop learning machine parameters and demonstrate the effectiveness of this process. This paper will also demonstrate that the applied EP process will reduce the amount of time required to configure a learning machine for those data sets studied, while developing optimal learning parameters with minimal user intervention. Specifically, this research has demonstrated, using the Duke mammogram data set, that SVMs derived using this modified EP process improved the specificity by a significant 45.3% at 100% sensitivity (missing no cancers) as well as improving the specificity by 17.5% at 95% sensitivity (missing 5% of the cancers) when compared to the performance of SVMs whose parameters were computed using the standard iterative method. The practical consequence of these results is that many women, who currently have false positive diagnoses resulting from the application of existing methods, will no longer be required to undergo biopsy with the resultant cost, morbidity and physical disfigurement that frequently results from these procedures. This approach, in addition, may also be used in linear separable; linear, non-separable; and nonlinear, non-separable environments.

Authors
Jr, WHL; Lo, JY; Velázquez, R
MLA Citation
Jr, WHL, Lo, JY, and Velázquez, R. "Using evolutionary programming to configure support vector machines for the diagnosis of breast cancer." Intelligent Engineering Systems Through Artificial Neural Networks 12 (2002): 249-254.
Source
scival
Published In
Intelligent Engineering Systems Through Artificial Neural Networks
Volume
12
Publish Date
2002
Start Page
249
End Page
254

Computerized classification of suspicious regions in chest radiographs using subregion Hotelling observers.

We propose to investigate the use of subregion Hotelling observers (SRHOs) in conjunction with perceptrons for the computerized classification of suspicious regions in chest radiographs for being nodules requiring follow up. Previously, 239 regions of interest (ROIs), each containing a suspicious lesion with proven classification, were collected. We chose to investigate the use of SRHOs as part of a multilayer classifier to determine the presence of a nodule. Each SRHO incorporates information about signal, background, and noise correlation for classification. For this study, 225 separate Hotelling observers were set up in a grid across each ROI. Each separate observer discriminates an 8 by 8 pixel area. A round robin sampling scheme was used to generate the 225 features, where each feature is the output of the individual observers. These features were then rank ordered by the magnitude of the weights of a perceptron. Once rank ordered, subsets of increasing number of features were selected to be used in another perceptron. This perceptron was trained to minimize mean squared error and the output was a continuous variable representing the likelihood of the region being a nodule. Performance was evaluated by receiver operating characteristic (ROC) analysis and reported as the area under the curve (Az). The classifier was optimized by adding additional features until the Az declined. The optimized subset of observers then were combined using a third perceptron. A subset of 80 features was selected which gave an Az of 0.972. Additionally, at 98.6% sensitivity, the classifier had a specificity of 71.3% and increased the positive predictive value from 60.7% to 84.1 %. Preliminary results suggest that using SRHOs in combination with perceptrons can provide a successful classification scheme for pulmonary nodules. This approach could be incorporated into a larger computer aided detection system for decreasing false positives.

Authors
Baydush, AH; Catarious, DM; Lo, JY; Abbey, CK; Floyd, CE
MLA Citation
Baydush, AH, Catarious, DM, Lo, JY, Abbey, CK, and Floyd, CE. "Computerized classification of suspicious regions in chest radiographs using subregion Hotelling observers." Med Phys 28.12 (December 2001): 2403-2409.
PMID
11797942
Source
pubmed
Published In
Medical physics
Volume
28
Issue
12
Publish Date
2001
Start Page
2403
End Page
2409
DOI
10.1118/1.1420402

Computer-aided detection of lung nodules in chest radiographs using sub-region Hotelling observers

Authors
Baydush, AH; Catarious, DM; Lo, JY; Abbey, CK; Floyd, CE
MLA Citation
Baydush, AH, Catarious, DM, Lo, JY, Abbey, CK, and Floyd, CE. "Computer-aided detection of lung nodules in chest radiographs using sub-region Hotelling observers." November 2001.
Source
wos-lite
Published In
Radiology
Volume
221
Publish Date
2001
Start Page
548
End Page
548

A neural network approach to breast cancer diagnosis as a constraint satisfaction problem.

A constraint satisfaction neural network (CSNN) approach is proposed for breast cancer diagnosis using mammographic and patient history findings. Initially, the diagnostic decision to biopsy was formulated as a constraint satisfaction problem. Then, an associative memory type neural network was applied to solve the problem. The proposed network has a flexible, nonhierarchical architecture that allows it to operate not only as a predictive tool but also as an analysis tool for knowledge discovery of association rules. The CSNN was developed and evaluated using a database of 500 nonpalpable breast lesions with definitive histopathological diagnosis. The CSNN diagnostic performance was evaluated using receiver operating characteristic analysis (ROC). The results of the study showed that the CSNN ROC area index was 0.84+/-0.02. The CSNN predictive performance is competitive with that achieved by experienced radiologists and backpropagation artificial neural networks (BP-ANNs) presented before. Furthermore, the study illustrates how CSNN can be used as a knowledge discovery tool overcoming some of the well-known limitations of BP-ANNs.

Authors
Tourassi, GD; Markey, MK; Lo, JY; Floyd, CE
MLA Citation
Tourassi, GD, Markey, MK, Lo, JY, and Floyd, CE. "A neural network approach to breast cancer diagnosis as a constraint satisfaction problem." Med Phys 28.5 (May 2001): 804-811.
PMID
11393476
Source
pubmed
Published In
Medical physics
Volume
28
Issue
5
Publish Date
2001
Start Page
804
End Page
811
DOI
10.1118/1.1367861

New results in breast cancer classification obtained from an evolutionary computation/adaptive boosting hybrid using mammogram and history data

© 2001 IEEE.A new neural network technology was developed to improve the diagnosis of breast cancer using mammogram findings. The paradigm, adaptive boosting (AB), uses a markedly different theory in solving the computational intelligence (CI) problem. AB, a new machine learning paradigm, focuses on finding weak learning algorithm(s) that initially need to provide slightly better than "random" performance (i.e., approximately 55%) when processing a mammogram training set. By successive development of additional architectures (using the mammogram training set), the adaptive boosting process improves performance of the basic evolutionary programming derived neural network architectures. The results of these several EP-derived hybrid architectures are then intelligently combined and tested using a similar validation mammogram data set. Optimization, focused on improving specificity and positive predictive value at very high sensitivities, with an analysis of the performance of the hybrid would be most meaningful. Using the DUKE mammogram database of 500 biopsy proven samples, this hybrid, on average, was able to achieve (under statistical 5-fold cross-validation) a specificity of 48.3% and a positive predictive value (PPV) of 51.8% while maintaining 100% sensitivity. At 97% sensitivity, a specificity of 56.6% and a PPV of 55.8% were obtained.

Authors
Land, WH; Masters, T; Lo, JY; McKee, DW; Anderson, FR
MLA Citation
Land, WH, Masters, T, Lo, JY, McKee, DW, and Anderson, FR. "New results in breast cancer classification obtained from an evolutionary computation/adaptive boosting hybrid using mammogram and history data." January 1, 2001.
Source
scopus
Published In
SMCia 2001 - Proceedings of the 2001 IEEE Mountain Workshop on Soft Computing in Industrial Applications
Publish Date
2001
Start Page
47
End Page
52
DOI
10.1109/SMCIA.2001.936727

Application of adaptive boosting to EP-derived multi-layer feedforward neural networks (MLFN) to improve benign/malignant breast cancer classification

A new neural network technology was developed for improving the benign/malignant diagnosis of breast cancer using mammogram findings. A new paradigm, Adaptive Boosting (AB), uses a markedly different theory in solutioning Computational Intelligence (CI) problems. AB, a new machine learning paradigm, focuses on finding weak learning algorithm(s) that initially need to provide slightly better than "random" performance (i.e., approximately 55%) when processing a mammogram training set. Then, by successive development of additional architectures (using the mammogram training set), the adaptive boosting process improves the performance of the basic Evolutionary Programming derived neural network architectures. The results of these several EP-derived hybrid architectures are then intelligently combined and tested using a similar validation mammogram data set. Optimization focused on improving specificity and positive predictive value at very high sensitivities, where an analysis of the performance of the hybrid would be most meaningful. Using the DUKE mammogram database of 500 biopsy proven samples, on average this hybrid was able to achieve (under statistical 5-fold cross-validation) a specificity of 48.3% and a positive predictive value (PPV) of 51.8% while maintaining 100% sensitivity. At 97% sensitivity, a specificity of 56.6% and a PPV of 55.8% were obtained.

Authors
Jr, LWH; Masters, T; Lo, JY; McKee, DW
MLA Citation
Jr, LWH, Masters, T, Lo, JY, and McKee, DW. "Application of adaptive boosting to EP-derived multi-layer feedforward neural networks (MLFN) to improve benign/malignant breast cancer classification." Proceedings of SPIE - The International Society for Optical Engineering 4322.3 (2001): 1717-1724.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
4322
Issue
3
Publish Date
2001
Start Page
1717
End Page
1724
DOI
10.1117/12.431058

Application of evolutionary computation and neural network hybrids for breast cancer classification using mammogram and history data

Mammography is the modality of choice for the early detection of breast cancer, primarily because of its sensitivity to the detection of breast cancer. However, because of its high rate of false positive predictions, a large number of biopsies of benign lesions result. This paper explores the use and evaluates the performance of two neural network hybrids as an aid to radiologists in avoiding biopsies of these benign lesions. These hybrids provide the potential to improve both the sensitivity and specificity of breast cancer diagnosis. The first hybrid, the Generalized Regression Neural Network (GRNN) Oracle, focuses on improving the performance output of a set of learning algorithms that operate and are accurate over the entire (defined) learning space. The second hybrid, an Evolutionary Programming (EP)/Adaptive Boosting (AB) based hybrid, intelligently combines the outputs from an iteratively called 'weak" learning algorithm (one which performs at least slightly better than random guessing) in or der to "boost" the performance of the weak learner. The second part of this paper discusses modifications to improve the EP/AB hybrid's performance, and further evaluates how the use of the EP/AB hybrid may obviate biopsies of benign lesions (as compared to an EP only classification system), given the requirement of missing few if any cancers.

Authors
Jr, LWH; Masters, T; Lo, JY; McKee, DW
MLA Citation
Jr, LWH, Masters, T, Lo, JY, and McKee, DW. "Application of evolutionary computation and neural network hybrids for breast cancer classification using mammogram and history data." Proceedings of the IEEE Conference on Evolutionary Computation, ICEC 2 (2001): 1147-1154.
Source
scival
Published In
Proceedings of the IEEE Conference on Evolutionary Computation, ICEC
Volume
2
Publish Date
2001
Start Page
1147
End Page
1154

Case-based reasoning computer algorithm that uses mammographic findings for breast biopsy decisions.

OBJECTIVE: We present case-based reasoning computer software developed from mammographic findings to provide support for the clinical decision to perform biopsy of the breast. SUBJECTS AND METHODS: The case-based reasoning system is designed to support the decision to perform biopsy in those patients who have suspicious findings on diagnostic mammography. Currently, between 66% and 90% of biopsies are performed on benign lesions. Our system is designed to help decrease the number of benign biopsies without missing malignancies. Clinicians interpret the mammograms using a standard reporting lexicon. The case-based reasoning system compares these findings with a database of cases with known outcomes (from biopsy) and returns the fraction of similar cases that were malignant. This malignancy fraction is an intuitive response that the clinician can then consider when making the decision regarding biopsy. RESULTS: The system was evaluated using a round-robin sampling scheme and performed with an area under the receiver operating characteristic curve of 0.83, comparable with the performance of a neural network model. If only the cases returning a malignancy fraction of greater than a threshold of 0.10 are sent to biopsy, no malignancies would be missed, and the number of benign biopsies would be decreased by 25%. At a threshold of 0.21, 98%, of the malignancies would be biopsied, and the number of benign biopsies would be decreased by 41%. CONCLUSION: This preliminary investigation indicates that the case-based reasoning approach to computer-aided diagnosis has the potential to improve the accuracy of breast cancer diagnosis on mammography.

Authors
Floyd, CE; Lo, JY; Tourassi, GD
MLA Citation
Floyd, CE, Lo, JY, and Tourassi, GD. "Case-based reasoning computer algorithm that uses mammographic findings for breast biopsy decisions." AJR Am J Roentgenol 175.5 (November 2000): 1347-1352.
PMID
11044039
Source
pubmed
Published In
AJR. American journal of roentgenology
Volume
175
Issue
5
Publish Date
2000
Start Page
1347
End Page
1352
DOI
10.2214/ajr.175.5.1751347

Segmentation of suspicious clustered microcalcifications in mammograms.

We have developed a multistage computer-aided diagnosis (CAD) scheme for the automated segmentation of suspicious microcalcification clusters in digital mammograms. The scheme consisted of three main processing steps. First, the breast region was segmented and its high-frequency content was enhanced using unsharp masking. In the second step, individual microcalcifications were segmented using local histogram analysis on overlapping subimages. For this step, eight histogram features were extracted for each subimage and were used as input to a fuzzy rule-based classifier that identified subimages containing microcalcifications and assigned the appropriate thresholds to segment any microcalcifications within them. The final step clustered the segmented microcalcifications and extracted the following features for each cluster: the number of microcalcifications, the average distance between microcalcifications, and the average number of times pixels in the cluster were segmented in the second step. Fuzzy logic rules incorporating the cluster features were designed to remove nonsuspicious clusters, defined as those with typically benign characteristics. A database of 98 images, with 48 images containing one or more microcalcification clusters, provided training and testing sets to optimize the parameters and evaluate the CAD scheme, respectively. The results showed a true positive rate of 93.2% and an average of 0.73 false positive clusters per image. A comparison of our results with other reported segmentation results on the same database showed comparable sensitivity and at the same time an improved false positive rate. The performance of the CAD scheme is encouraging for its use as an automatic tool for efficient and accurate diagnosis of breast cancer.

Authors
Gavrielides, MA; Lo, JY; Vargas-Voracek, R; Floyd, CE
MLA Citation
Gavrielides, MA, Lo, JY, Vargas-Voracek, R, and Floyd, CE. "Segmentation of suspicious clustered microcalcifications in mammograms." Med Phys 27.1 (January 2000): 13-22.
PMID
10659733
Source
pubmed
Published In
Medical physics
Volume
27
Issue
1
Publish Date
2000
Start Page
13
End Page
22
DOI
10.1118/1.598852

Application of a GRNN oracle to the intelligent combination of several breast cancer benign/malignant predictive paradigms

The General Regression Neural Network (GRNN) is well known to be an extremely effective prediction model in a wide variety of problems. It has been recently established that in many prediction problems, the results obtained by intelligently combining the outputs of several different prediction models are generally superior to the results obtained by using any one of the models. An overseer model that combines predictions from other independently trained prediction models is often called an oracle. This paper describes how the GRNN is modified to serve as a powerful oracle for combining decisions from four different breast cancer benign/malignant prediction models using mammogram data. Specifically, the GRNN oracle combines decisions from an evolutionary programming derived neural network, a probabilistic neural network, a fully-interconnected three-layer, feed-forward, error backpropagation network, and a linear discriminant analysis model. In all experiments conducted, the oracle consistently provided superior benign/malignant classification discrimination as measured by the receiver operator characteristic curve Az index values

Authors
Land, WH; Jr, MT; Lo, JY
MLA Citation
Land, WH, Jr, MT, and Lo, JY. "Application of a GRNN oracle to the intelligent combination of several breast cancer benign/malignant predictive paradigms." Proc. SPIE - Int. Soc. Opt. Eng. (USA) 3979 (2000): 77-85. (Academic Article)
Source
manual
Published In
Proc. SPIE - Int. Soc. Opt. Eng. (USA)
Volume
3979
Publish Date
2000
Start Page
77
End Page
85

Use of a constraint satisfaction neural network for breast cancer diagnosis and dynamic scenarios simulation

Authors
Tourassi, GD; Floyd, CE; Lo, JY
MLA Citation
Tourassi, GD, Floyd, CE, and Lo, JY. "Use of a constraint satisfaction neural network for breast cancer diagnosis and dynamic scenarios simulation." MEDICAL IMAGING 2000: IMAGE PROCESSING, PTS 1 AND 2 3979 (2000): 46-54.
Source
wos-lite
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3979
Publish Date
2000
Start Page
46
End Page
54
DOI
10.1117/12.387680

Evolutionary programming technique for reducing complexity of artificial neural networks for breast cancer diagnosis

Authors
Lo, JY; Land, WH; Morrison, CT
MLA Citation
Lo, JY, Land, WH, and Morrison, CT. "Evolutionary programming technique for reducing complexity of artificial neural networks for breast cancer diagnosis." MEDICAL IMAGING 2000: IMAGE PROCESSING, PTS 1 AND 2 3979 (2000): 153-158.
Source
wos-lite
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3979
Publish Date
2000
Start Page
153
End Page
158
DOI
10.1117/12.387635

Application of a GRNN ORACLE to the intelligent combination of several breast cancer benign/malignant predictive paradigms

The General Regression Neural Network (GRNN) is well known to be an extremely effective prediction model in a wide variety of problems. It has been recently established that in many prediction problems, the results obtained by intelligently combining the outputs of several different prediction models are generally superior to the results obtained by using any one of the models. An overseer model that combines predictions from other independently trained prediction models is often called an oracle. This paper describes how the GRNN is modified to serve as a powerful oracle for combining decisions from four different breast cancer benign/malignant prediction models using mammogram data. Specifically, the GRNN oracle combines decisions from an evolutionary programming derived neural network, a probabilistic neural network, a fully-interconnected three-layer, feed-forward, error backpropagation network, and a linear discriminant analysis model. In all experiments conducted, the oracle consistently provided superior benign/malignant classification discrimination as measured by the receiver operator characteristic curve Az index values.

Authors
Jr, WHL; Masters, T; Lo, JY
MLA Citation
Jr, WHL, Masters, T, and Lo, JY. "Application of a GRNN ORACLE to the intelligent combination of several breast cancer benign/malignant predictive paradigms." Proceedings of SPIE - The International Society for Optical Engineering 3979 (2000): I/--.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3979
Publish Date
2000
Start Page
I/-

Evolutionary programming technique for reducing complexity of artificial neural networks for breast cancer diagnosis

An evolutionary programming (EP) technique was investigated to reduce the complexity of artificial neural network (ANN) models that predict the outcome of mammography-induced breast biopsy. By combining input variables consisting of mammography lesion descriptors and patient history data, the ANN predicted whether the lesion was benign or malignant, which may aide in reducing the number of unnecessary benign biopsies and thus the cost of mammography screening of breast cancer. The EP has the ability to optimize the ANN both structurally and parametrically. An EP was partially optimized using a data set of 882 biopsy-proven cases from Duke University Medical Center. Although many different architectures were evolved, the best were often perceptrons with no hidden nodes. A rank ordering of the inputs was performed using twenty independent EP runs. This confirmed the predictive value of the mass margin and patient age variables, and revealed the unexpected usefulness of the history of previous breast cancer. Further work is required to improve the performance of the EP over all cases in general and calcification cases in particular.

Authors
Lo, JY; Land, WH; Morrison, CT
MLA Citation
Lo, JY, Land, WH, and Morrison, CT. "Evolutionary programming technique for reducing complexity of artificial neural networks for breast cancer diagnosis." Proceedings of SPIE - The International Society for Optical Engineering 3979 (2000): I/--.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3979
Publish Date
2000
Start Page
I/-

Use of a constraint satisfaction neural network for breast cancer diagnosis and dynamic scenarios simulation

A constraint satisfaction neural network (CSNN) has been developed for breast cancer diagnosis from mammographic and clinical findings. CSNN is a circuit network aiming to maximize the activation of its nodes given the constraints existing among them. The constraints are built into the network weights. An autoassociative backpropagation (auto-BP) learning scheme is initially used to determine the CSNN weights. During the training phase, the auto-BP learns to map any given pattern to itself. During the testing phase, the CSNN is applied to new cases. The CSNN weights remain fixed (as determined by auto-BP) but the activation levels of the nodes are modified iteratively to optimize a goodness function. The medical findings act as the external inputs to the corresponding nodes. For every test case, CSNN tries to reconstruct the diagnosis nodes given the network constraints and the external inputs to the network. The activation levels achieved by the target nodes are used as decision variables for further analysis. Our CSNN was successfully applied on 500 patients with biopsy confirmed diagnosis. The CSNN was also used as an associative memory simulating dynamic scenarios for prototype analysis in our database.

Authors
Tourassi, GD; Jr, CEF; Lo, JY
MLA Citation
Tourassi, GD, Jr, CEF, and Lo, JY. "Use of a constraint satisfaction neural network for breast cancer diagnosis and dynamic scenarios simulation." Proceedings of SPIE - The International Society for Optical Engineering 3979 (2000): I/--.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3979
Publish Date
2000
Start Page
I/-

Application of a new Evolutionary Programming/Adaptive Boosting hybrid to breast cancer diagnosis

A new Evolutionary Programming/Adaptive Boosting (EP/AB) neural network hybrid was investigated to measure the hybrid performance improvement as obtained when using an EP-only derived neural network as a baseline. By combining input variables consisting of mammography lesion descriptors and patient history data, the hybrid predicted whether the lesion was benign or malignant, which may aid in reducing the number of unnecessary biopsies and thus the cost of mammography screening of breast cancer. The EP process as well as the hybrid was optimized using a data set of 500 biopsy-proven cases from Duke University Medical Center. Results showed that the hybrid provided a 15-20% classification performance improvement as measured by the ROC Az index when compared to a non-optimized EP derived architecture.

Authors
Jr, WL; Masters, T; Lo, J
MLA Citation
Jr, WL, Masters, T, and Lo, J. "Application of a new Evolutionary Programming/Adaptive Boosting hybrid to breast cancer diagnosis." Proceedings of the IEEE Conference on Evolutionary Computation, ICEC 2 (2000): 1436-1442.
Source
scival
Published In
Proceedings of the IEEE Conference on Evolutionary Computation, ICEC
Volume
2
Publish Date
2000
Start Page
1436
End Page
1442

A neural network to predict symptomatic lung injury.

A nonlinear neural network that simultaneously uses pre-radiotherapy (RT) biological and physical data was developed to predict symptomatic lung injury. The input data were pre-RT pulmonary function, three-dimensional treatment plan doses and demographics. The output was a single value between 0 (asymptomatic) and 1 (symptomatic) to predict the likelihood that a particular patient would become symptomatic. The network was trained on data from 97 patients for 400 iterations with the goal to minimize the mean-squared error. Statistical analysis was performed on the resulting network to determine the model's accuracy. Results from the neural network were compared with those given by traditional linear discriminate analysis and the dose-volume histogram reduction (DVHR) scheme of Kutcher. Receiver-operator characteristic (ROC) analysis was performed on the resulting network which had Az = 0.833 +/- 0.04. (Az is the area under the ROC curve.) Linear discriminate multivariate analysis yielded an Az = 0.813 +/- 0.06. The DVHR method had Az = 0.521 +/- 0.08. The network was also used to rank the significance of the input variables. Future studies will be conducted to improve network accuracy and to include functional imaging data.

Authors
Munley, MT; Lo, JY; Sibley, GS; Bentel, GC; Anscher, MS; Marks, LB
MLA Citation
Munley, MT, Lo, JY, Sibley, GS, Bentel, GC, Anscher, MS, and Marks, LB. "A neural network to predict symptomatic lung injury." Phys Med Biol 44.9 (September 1999): 2241-2249.
PMID
10495118
Source
pubmed
Published In
Physics in Medicine and Biology
Volume
44
Issue
9
Publish Date
1999
Start Page
2241
End Page
2249

Application of artificial neural networks for diagnosis of breast cancer

We review four current projects pertaining to artificial neural network (ANN) models that merge radiologist-extracted findings to perform computer aided diagnosis (CADx) of breast cancer. These projects are: (1) prediction of breast lesion malignancy using mammographic findings; (2) classification of malignant lesions as in situ vs. invasive cancer; (3) prediction of breast mass malignancy using ultrasound findings; and (4) the evaluation of CADx models in a cross-institution study. These projects share in common the use of feedforward error backpropagation ANNs. Inputs to the ANNs are medical findings such as mammographic or ultrasound lesion descriptors and patient history data. The output is the biopsy outcome (benign vs. malignant, or in situ vs. invasive cancer) which is being predicted. All ANNs undergo supervised training using actual patient data. These ANN decision models may assist in the management of patients with breast lesions, such as by reducing the number of unnecessary surgical procedures and their associated cost. © 1999 IEEE.

Authors
Lo, JY; Floyd, CE
MLA Citation
Lo, JY, and Floyd, CE. "Application of artificial neural networks for diagnosis of breast cancer." Proceedings of the 1999 Congress on Evolutionary Computation, CEC 1999 3 (January 1, 1999): 1755-1759.
Source
scopus
Published In
Proceedings of the 1999 Congress on Evolutionary Computation, CEC 1999
Volume
3
Publish Date
1999
Start Page
1755
End Page
1759
DOI
10.1109/CEC.1999.785486

Effect of patient history data on the prediction of breast cancer from mammographic findings with artificial neural networks.

RATIONALE AND OBJECTIVES: The authors evaluated the contribution of medical history data to the prediction of breast cancer with artificial neural network (ANN) models based on mammographic findings. MATERIALS AND METHODS: Three ANNs were developed: The first used 10 Breast Imaging Reporting and Data System (BI-RADS) variables; the second, the BI-RADS variables plus patient age; the third, the BI-RADS variables, patient age, and seven other history variables, for a total of 18 inputs. Performance of the ANNs and the original radiologist's impression were evaluated with five metrics: receiver operating characteristic area index (Az); specificity at given sensitivities of 100%, 98%, and 95%; and positive predictive value. RESULTS: All three ANNs consistently outperformed the radiologist's impression over all five performance metrics. The patient-age variable was particularly valuable. Adding the age variable to the basic ANN model, which used only the BI-RADS findings, significantly improved Az (P = .028). In fact, replacing all history data with just the age variable resulted in virtually no changes for Az or specificity at 98% sensitivity (P = .324 and P = .410, respectively). CONCLUSION: Patient age was an important variable for the prediction of breast cancer from mammographic findings with the ANNs. For this data set, all history data could be replaced with age alone.

Authors
Lo, JY; Baker, JA; Kornguth, PJ; Floyd, CE
MLA Citation
Lo, JY, Baker, JA, Kornguth, PJ, and Floyd, CE. "Effect of patient history data on the prediction of breast cancer from mammographic findings with artificial neural networks." Acad Radiol 6.1 (January 1999): 10-15.
PMID
9891147
Source
pubmed
Published In
Academic Radiology
Volume
6
Issue
1
Publish Date
1999
Start Page
10
End Page
15

Application of a GRNN oracle to the intelligent combination of several breast cancer benign/malignant predictive paradigms

The General Regression Neural Network (GRNN) is well known to be an extremely effective prediction model in a wide variety of problems. It has been recently established that in many prediction problems, the results obtained by intelligently combining the outputs of several different prediction models are generally superior to the results obtained by using any one of the models. An overseer model that combines predictions from other independently trained prediction models is often called an oracle. This paper describes how the GRNN is modified to serve as a powerful oracle for combining decisions from four different breast cancer benign/malignant prediction models using mammogram data. In all experiments conducted, the oracle consistently provided superior benign/malignant classification discrimination as measured by the receiver operator characteristic curve Az index values.

Authors
Jr, WHL; Masters, T; Morrison, CT; Lo, JY
MLA Citation
Jr, WHL, Masters, T, Morrison, CT, and Lo, JY. "Application of a GRNN oracle to the intelligent combination of several breast cancer benign/malignant predictive paradigms." Intelligent Engineering Systems Through Artificial Neural Networks 9 (1999): 803-808.
Source
scival
Published In
Intelligent Engineering Systems Through Artificial Neural Networks
Volume
9
Publish Date
1999
Start Page
803
End Page
808

Case-based reasoning as a computer aid to diagnosis

A Case-Based Reasoning (CBR) system has been developed to predict the outcome of excisional biopsy from mammographic findings. CBR is implemented by comparing the current case to all previous cases and examining the outcomes for those previous cases that match the current case. Patients from breast screening who have suspicious findings on their diagnostic mammogram, are candidates for biopsy. The false positive rate for the decision to biopsy is currently between 66% and 90%. The CBR system is designed to support the decision to biopsy. The mammograms are read by clinicians using a standard reporting lexicon (BI-RADSTM). These findings are compared to a database of findings from cases with known outcomes (from biopsy). The fraction of similar cases that were malignant is returned. The clinician can then consider this result when making the decision regarding biopsy. The system was evaluated using round-robin sampling scheme and performed with a Receiver Operating Characteristic (ROC) area of 0.77.

Authors
Jr, CEF; Lo, JY; Tourassi, GD
MLA Citation
Jr, CEF, Lo, JY, and Tourassi, GD. "Case-based reasoning as a computer aid to diagnosis." Proceedings of SPIE - The International Society for Optical Engineering 3661.I (1999): 486-489.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
3661
Issue
I
Publish Date
1999
Start Page
486
End Page
489

Application of evolutionary programming and probabilistic neural networks to breast cancer diagnosis

Two novel artificial neural network techniques, evolutionary programming (EP) and probabilistic neural networks (PNN), were applied to the problem of breast cancer diagnosis. The EP is a stochastic optimization technique with the ability to mutate both network connections and weight values. The PNN has the ability to produce optimal Bayesian decision making given sufficient training data. Both techniques offer potential improvements over the well-studied, classic backpropagation networks. Preliminary performances of these new techniques were comparable to but slightly worse than the classic networks. In on-going work, these new techniques will be optimized further and should produce results greater than or equal to the classic networks, but with more information content and confidence.

Authors
Lo, JY; Land, WH; Morrison, CT
MLA Citation
Lo, JY, Land, WH, and Morrison, CT. "Application of evolutionary programming and probabilistic neural networks to breast cancer diagnosis." Proceedings of the International Joint Conference on Neural Networks 5 (1999): 3712-3716.
Source
scival
Published In
Proceedings of the International Joint Conference on Neural Networks
Volume
5
Publish Date
1999
Start Page
3712
End Page
3716

Constraint Satisfaction Neural Network for medical diagnosis

This objective of this study was to explore how a Constraint Satisfaction Neural Network (CSNN) can be used for medical diagnostic tasks. The study is based on a database of 500 patients who underwent breast biopsy at Duke University Medical Center due to suspicious mammographic findings. A CSNN was developed and evaluated to predict the biopsy result from the patient's mammographic findings. The diagnostic performance of the CSNN network was compared to a traditional backpropagation (BP) neural network and a case-based-reasoning (CBR) algorithm by means of Receiver Operating characteristics (ROC) analysis. This study demonstrates (i) how CSNNs can be applied for medical diagnostic tasks and, (ii) how they can be utilized to extract meaningful clinical information regarding underlying relationships among medical findings and associated diagnoses.

Authors
Tourassi, GD; Jr, CEF; Lo, JY
MLA Citation
Tourassi, GD, Jr, CEF, and Lo, JY. "Constraint Satisfaction Neural Network for medical diagnosis." Proceedings of the International Joint Conference on Neural Networks 5 (1999): 3632-3635.
Source
scival
Published In
Proceedings of the International Joint Conference on Neural Networks
Volume
5
Publish Date
1999
Start Page
3632
End Page
3635

Prediction of breast biopsy outcomes from mammographic findings

Authors
Floyd, CE; Lo, JY; Baker, JA
MLA Citation
Floyd, CE, Lo, JY, and Baker, JA. "Prediction of breast biopsy outcomes from mammographic findings." COMPUTER-AIDED DIAGNOSIS IN MEDICAL IMAGING 1182 (1999): 193-200.
Source
wos-lite
Published In
International Congress Series
Volume
1182
Publish Date
1999
Start Page
193
End Page
200

Computer-aided diagnosis of breast cancer

Authors
Lo, JY; Floyd, CE
MLA Citation
Lo, JY, and Floyd, CE. "Computer-aided diagnosis of breast cancer." COMPUTER-AIDED DIAGNOSIS IN MEDICAL IMAGING 1182 (1999): 221-225.
Source
wos-lite
Published In
International Congress Series
Volume
1182
Publish Date
1999
Start Page
221
End Page
225

Predictive model for the diagnosis of intraabdominal abscess.

RATIONALE AND OBJECTIVES: The authors investigated the use of an artificial neural network (ANN) to aid in the diagnosis of intraabdominal abscess. MATERIALS AND METHODS: An ANN was constructed based on data from 140 patients who underwent abdominal and pelvic computed tomography (CT) between January and December 1995. Input nodes included data from clinical history, physical examination, laboratory investigation, and radiographic study. The ANN was trained and tested on data from all 140 cases by using a round-robin method and was compared with linear discriminate analysis. A receiver operating characteristic curve was generated to evaluate both predictive models. RESULTS: CT examinations in 50 cases were positive for abscess. This finding was confirmed by means of laboratory culture of aspirations from CT-guided percutaneous drainage in 38 patients, ultrasound-guided percutaneous drainage in five patients, surgery in five patients, and characteristic appearance on CT scans without aspiration in two patients. CT scans in 90 cases were negative for abscess. The sensitivity and specificity of the ANN in predicting the presence of intraabdominal abscess were 90% and 51%, respectively. Receiver operating characteristic analysis showed no statistically significant difference in performance between the two predictive models. CONCLUSION: The ANN is a useful tool for determining whether an intraabdominal abscess is present. It can be used to set priorities for CT examinations in order to expedite treatment in patients believed to be more likely to have an abscess.

Authors
Freed, KS; Lo, JY; Baker, JA; Floyd, CE; Low, VH; Seabourn, JT; Nelson, RC
MLA Citation
Freed, KS, Lo, JY, Baker, JA, Floyd, CE, Low, VH, Seabourn, JT, and Nelson, RC. "Predictive model for the diagnosis of intraabdominal abscess." Acad Radiol 5.7 (July 1998): 473-479.
PMID
9653463
Source
pubmed
Published In
Academic Radiology
Volume
5
Issue
7
Publish Date
1998
Start Page
473
End Page
479

QoS middleware for Internet multimedia streaming

In real-time multimedia streaming, some control mechanism to maintain the quality of service (QoS) in data transfer, which includes not only network-level QoS guarantee by resource reservation but also adaptive control by end hosts according to the network conditions is necessary. NEC has been developing a middleware that will provide a framework of these QoS control functions by using Internet standard protocols such as RSVP (resource reservation protocol), RTP (real-time transport protocol) and RTSP (real-time streaming protocol). This paper presents the architecture of the QoS middleware, QoS control mechanism and the implemented system

Authors
Sakamoto, H; Lo, JYD; Nishida, T
MLA Citation
Sakamoto, H, Lo, JYD, and Nishida, T. "QoS middleware for Internet multimedia streaming." NEC Tech. J. (Japan) 51.8 (1998): 35-40. (Academic Article)
Source
manual
Published In
NEC Tech. J. (Japan)
Volume
51
Issue
8
Publish Date
1998
Start Page
35
End Page
40

Predicting breast cancer invasion with artificial neural networks on the basis of mammographic features.

PURPOSE: To evaluate whether an artificial neural network (ANN) can predict breast cancer invasion on the basis of readily available medical findings (ie, mammographic findings classified according to the American College of Radiology Breast Imaging Reporting and Data System and patient age). MATERIALS AND METHODS: In 254 adult patients, 266 lesions that had been sampled at biopsy were randomly selected for the study. There were 96 malignant and 170 benign lesions. On the basis of nine mammographic findings and patient age, a three-layer backpropagation network was developed to predict whether the malignant lesions were in situ or invasive. RESULTS: The ANN predicted invasion among malignant lesions with an area under the receiver operating characteristic curve (Az) of .91 +/- .03. It correctly identified all 28 in situ cancers (specificity, 100%) and 48 of 68 invasive cancers (sensitivity, 71%). CONCLUSION: The ANN used mammographic features and patient age to accurately classify invasion among breast cancers, information that was previously available only by means of biopsy. This knowledge may assist in surgical planning and may help reduce the cost and morbidity of unnecessary biopsy.

Authors
Lo, JY; Baker, JA; Kornguth, PJ; Iglehart, JD; Floyd, CE
MLA Citation
Lo, JY, Baker, JA, Kornguth, PJ, Iglehart, JD, and Floyd, CE. "Predicting breast cancer invasion with artificial neural networks on the basis of mammographic features." Radiology 203.1 (April 1997): 159-163.
PMID
9122385
Source
pubmed
Published In
Radiology
Volume
203
Issue
1
Publish Date
1997
Start Page
159
End Page
163
DOI
10.1148/radiology.203.1.9122385

Self-organizing maps for analyzing mammographic findings

The purpose of this study is to analyze mammographic findings using self-organizing map (SOM) artificial neural networks. Using two findings of patient age and mass margin extracted by radiologists, self-organizing maps were developed to analyze both the distribution and topology of the input findings. These results can help to explain the underlying nature of mammographic findings data, which may in turn help radiologists to improve breast cancer diagnosis and assist in the development of other neural networks.

Authors
Lo, JY; Jr, CEF
MLA Citation
Lo, JY, and Jr, CEF. "Self-organizing maps for analyzing mammographic findings." IEEE International Conference on Neural Networks - Conference Proceedings 4 (1997): 2472-2474.
Source
scival
Published In
IEEE International Conference on Neural Networks - Conference Proceedings
Volume
4
Publish Date
1997
Start Page
2472
End Page
2474

Diffuse nodular lung disease on chest radiographs: a pilot study of characterization by fractal dimension.

OBJECTIVE: We present a computer-aided diagnostic technique for identifying nodular interstitial lung disease on chest radiographs. The fractal dimension was used as a numerical measure of image texture on digital chest radiographs to distinguish patients with normal lung from those with a diffuse nodular interstitial abnormality. MATERIALS AND METHODS: Twenty digitized chest radiographs were classified as normal (n = 10) or as containing diffuse nodular abnormality (n = 10) on the basis of readings assigned according to the classification of the International Labour Organization. Regions of interest (ROIs) measuring 1.28 cm2 were selected from the intercostal spaces of these radiographs. The fractal dimension of these ROIs was estimated by power spectrum analysis. The cases were not subtle. RESULTS: The fractal dimension provided statistically significant discrimination between normal parenchyma and nodular interstitial lung disease. The area under the receiver operating characteristic curve was 0.90 (+/- 0.02). One operating point provides sensitivity of 88% with a specificity of 80%. CONCLUSION: The fractal dimension can provide a measure of lung parenchymal texture and shows promise as an element of computer-aided diagnosis, characterization, and follow-up of interstitial lung disease.

Authors
Floyd, CE; Patz, EF; Lo, JY; Vittitoe, NF; Stambaugh, LE
MLA Citation
Floyd, CE, Patz, EF, Lo, JY, Vittitoe, NF, and Stambaugh, LE. "Diffuse nodular lung disease on chest radiographs: a pilot study of characterization by fractal dimension." AJR Am J Roentgenol 167.5 (November 1996): 1185-1187.
PMID
8911177
Source
pubmed
Published In
AJR. American journal of roentgenology
Volume
167
Issue
5
Publish Date
1996
Start Page
1185
End Page
1187
DOI
10.2214/ajr.167.5.8911177

Artificial neural network: improving the quality of breast biopsy recommendations.

PURPOSE: To evaluate the performance and inter- and intraobserver variability of an artificial neural network (ANN) for predicting breast biopsy outcome. MATERIALS AND METHODS: Five radiologists described 60 mammographically detected lesions with the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) nomenclature. A previously programmed ANN used the BI-RADS descriptors and patient histories to predict biopsy results. ANN predictive performance was compared with the clinical decision to perform biopsy. Inter- and intraobserver variability of radiologists' interpretations and ANN predictions were evaluated with Cohen kappa analysis. RESULTS: The ANN maintained 100% sensitivity (23 of 23 cancers) while improving the positive predictive value of biopsy results from 38% (23 of 60 lesions) to between 58% (23 of 40 lesions) and 66% (23 of 35 lesions; P < .001). Interobserver variability for interpretation of the lesions was significantly reduced by the ANN (P < .001); there was no statistically significant effect on nearly perfect intraobserver reproducibility. CONCLUSION: Use of an ANN with radiologists' descriptions of abnormal findings may improve interpretation of mammographic abnormalities.

Authors
Baker, JA; Kornguth, PJ; Lo, JY; Floyd, CE
MLA Citation
Baker, JA, Kornguth, PJ, Lo, JY, and Floyd, CE. "Artificial neural network: improving the quality of breast biopsy recommendations." Radiology 198.1 (January 1996): 131-135.
PMID
8539365
Source
pubmed
Published In
Radiology
Volume
198
Issue
1
Publish Date
1996
Start Page
131
End Page
135
DOI
10.1148/radiology.198.1.8539365

Computer-aided diagnosis of mammography using an artificial neural network: Predicting the invasiveness of breast cancers from image features

The study aimed to develop an artificial neural network (ANN) for computer-aided diagnosis of mammography. Using 9 mammographic image features and patient age, the ANN predicted whether breast lesions were benign, invasive malignant, or noninvasive malignant. Given only 97 malignant patients, the 3-layer backpropagation ANN successfully predicted the invasiveness of those breast cancers, performing with Az of 0.88 ± 0.03. To determine more generalized clinical performance, a different ANN was developed using 266 consecutive patients (97 malignant, 169 benign). This ANN predicted whether those patients were benign or noninvasive malignant vs. invasive malignant with Az of 0.86 ± 0.03. This study is unique because it is the first to predict the invasiveness of breast cancers using mammographic features and age. This knowledge, which was previously available only through surgical biopsy, may assist in the planning of surgical procedures for patients with breast lesions, and may help reduce the cost and morbidity associated with unnecessary surgical biopsies.

Authors
Lo, JY; Kim, J; Baker, JA; Jr, CEF
MLA Citation
Lo, JY, Kim, J, Baker, JA, and Jr, CEF. "Computer-aided diagnosis of mammography using an artificial neural network: Predicting the invasiveness of breast cancers from image features." Proceedings of SPIE - The International Society for Optical Engineering 2710 (1996): 725-732.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
2710
Publish Date
1996
Start Page
725
End Page
732
DOI
10.1117/12.237977

Computer aided diagnosis in thoracic and mammographic radiology

There has been a significant effort in the radiology department at Duke University to develop computer aided diagnosis (CAD) systems. The goal of the development of these systems is to assist radiologists in interpreting radiographic images and findings. These efforts have encompassed: (1) detection and size quantification of cold lesions in nuclear medicine (SPECT) images; (2) diagnosis of acute pulmonary embolism from physicians reading of ventilation/perfusion scans and chest radiographs; (3) prediction of breast cancer malignancy from mammographers' readings of diagnostic mammograms; (4) detection of pulmonary nodules from digital chest radiographs; (5) classification of interstitial lung disease; and (6) characterization of pulmonary nodules. Details of two of these efforts are described here: (1) prediction of breast cancer malignancy from mammographers' readings of diagnostic mammograms and (2) detection of pulmonary nodules from digital chest radiographs

Authors
Jr, FCE; Lo, JY; Tourassi, GD; Baker, JA; Vitittoe, NF; Vargas-Vorack, R
MLA Citation
Jr, FCE, Lo, JY, Tourassi, GD, Baker, JA, Vitittoe, NF, and Vargas-Vorack, R. "Computer aided diagnosis in thoracic and mammographic radiology." Med. Imaging Technol. (Japan) 14.6 (1996): 629-634. (Academic Article)
Source
manual
Published In
Med. Imaging Technol. (Japan)
Volume
14
Issue
6
Publish Date
1996
Start Page
629
End Page
634

COMPUTER-AIDED DIAGNOSIS OF BREAST MASS MALIGNANCY WITH AUTOMATED FEATURE-EXTRACTION AND ARTIFICIAL NEURAL NETWORKS

Authors
LO, JY; BAYDUSH, AH; BAKER, JA; KORNGUTH, PJ; FLOYD, CE
MLA Citation
LO, JY, BAYDUSH, AH, BAKER, JA, KORNGUTH, PJ, and FLOYD, CE. "COMPUTER-AIDED DIAGNOSIS OF BREAST MASS MALIGNANCY WITH AUTOMATED FEATURE-EXTRACTION AND ARTIFICIAL NEURAL NETWORKS." November 1995.
Source
wos-lite
Published In
Radiology
Volume
197
Publish Date
1995
Start Page
425
End Page
425

Computer-aided diagnosis of breast cancer: artificial neural network approach for optimized merging of mammographic features.

RATIONALE AND OBJECTIVES: An artificial neural network (ANN) approach was developed for the computer-aided diagnosis of mammography using an optimally minimized number of input features. METHODS: A backpropagation ANN merged nine input features (age plus eight radiographic findings extracted by radiologists) to predict biopsy outcome as its output. The features were ranked, and more important ones were selected to produce an optimal subset of features. RESULTS: Given all nine features, the ANN performed with a receiver operator characteristic area under the curve (Az) of .95 +/- .01. Given only the four most important features, the ANN performed with an Az of .96 +/- .01. Although not significantly better than the ANN with all nine features, the ANN with the four optimized features was significantly better than expert radiologists' Az of .90 +/- .02 (p = .01). This four-feature ANN had a 95% sensitivity and an 81% specificity. For cases with calcifications, the radiologists' performance dropped to an Az of .85 +/- .04, whereas a specialized three-feature ANN performed significantly better with an Az of .95 +/- .02 (p = .02). CONCLUSION: Given only four input features, the ANN predicted biopsy outcome significantly better than did expert radiologists, who also had access to other radiographic and nonradiographic data. The reduced number of features would substantially decrease data entry efforts and potentially improve the ANN's general applicability.

Authors
Lo, JY; Baker, JA; Kornguth, PJ; Floyd, CE
MLA Citation
Lo, JY, Baker, JA, Kornguth, PJ, and Floyd, CE. "Computer-aided diagnosis of breast cancer: artificial neural network approach for optimized merging of mammographic features." Acad Radiol 2.10 (October 1995): 841-850.
PMID
9419649
Source
pubmed
Published In
Academic Radiology
Volume
2
Issue
10
Publish Date
1995
Start Page
841
End Page
850

Breast cancer: prediction with artificial neural network based on BI-RADS standardized lexicon.

PURPOSE: To determine if an artificial neural network (ANN) to categorize benign and malignant breast lesions can be standardized for use by all radiologists. MATERIALS AND METHODS: An ANN was constructed based on the standardized lexicon of the Breast Imaging Recording and Data System (BI-RADS) of the American College of Radiology. Eighteen inputs to the network included 10 BI-RADS lesion descriptors and eight input values from the patient's medical history. The network was trained and tested on 206 cases (133 benign, 73 malignant cases). Receiver operating characteristic curves for the network and radiologists were compared. RESULTS: At a specified output threshold, the ANN would have improved the positive predictive value (PPV) of biopsy from 35% to 61% with a relative sensitivity of 100%. At a fixed sensitivity of 95%, the specificity of the ANN (62%) was significantly greater than the specificity of radiologists (30%) (P < .01). CONCLUSION: The BI-RADS lexicon provides a standardized language between mammographers and an ANN that can improve the PPV of breast biopsy.

Authors
Baker, JA; Kornguth, PJ; Lo, JY; Williford, ME; Floyd, CE
MLA Citation
Baker, JA, Kornguth, PJ, Lo, JY, Williford, ME, and Floyd, CE. "Breast cancer: prediction with artificial neural network based on BI-RADS standardized lexicon." Radiology 196.3 (September 1995): 817-822.
PMID
7644649
Source
pubmed
Published In
Radiology
Volume
196
Issue
3
Publish Date
1995
Start Page
817
End Page
822
DOI
10.1148/radiology.196.3.7644649

Academic consortium for the evaluation of computer-aided diagnosis (CADx) in mammography

Computer aided diagnosis (CADx) is a promising technology for the detection of breast cancer in screening mammography. A number of different approaches have been developed for CADx research that have achieved significant levels of performance. Research teams now recognize the need for a careful and detailed evaluation study of approaches to accelerate the development of CADx, to make CADx more clinically relevant and to optimize the CADx algorithms based on unbiased evaluations. The results of such a comparative study may provide each of the participating teams with new insights into the optimization of their individual CADx algorithms. This consortium of experienced CADx researchers is working as a group to compare results of the algorithms and to optimize the performance of CADx algorithms by learning from each other. Each institution will be contributing an equal number of cases that will be collected under a standard protocol for case selection, truth determination, and data acquisition to establish a common and unbiased database for the evaluation study. An evaluation procedure for the comparison studies are being developed to analyze the results of individual algorithms for each of the test cases in the common database. Optimization of individual CADx algorithms can be made based on the comparison studies. The consortium effort is expected to accelerate the eventual clinical implementation of CADx algorithms at participating institutions.

Authors
Mun, SK; Freedman, MT; Wu, YC; Lo, BS; Jr, CEF; Lo, JY; Chan, H-P; Helvie, MA; Petrick, N; Sahiner, B; Wei, D; Chakraborty, DP; Clarke, LP; Kallergi, M; Clark, B; Kim, Y
MLA Citation
Mun, SK, Freedman, MT, Wu, YC, Lo, BS, Jr, CEF, Lo, JY, Chan, H-P, Helvie, MA, Petrick, N, Sahiner, B, Wei, D, Chakraborty, DP, Clarke, LP, Kallergi, M, Clark, B, and Kim, Y. "Academic consortium for the evaluation of computer-aided diagnosis (CADx) in mammography." Proceedings of SPIE - The International Society for Optical Engineering 2431 (1995): 442-446.
Source
scival
Published In
Proceedings of SPIE - The International Society for Optical Engineering
Volume
2431
Publish Date
1995
Start Page
442
End Page
446

COMPUTER-AIDED DIAGNOSIS OF MAMMOGRAMS USING AN ARTIFICIAL NEURAL NETWORK - MERGING OF STANDARDIZED INPUT FEATURES FROM THE ACR LEXICON

Authors
LO, JY; GRISSON, AT; FLOYD, CE; KORNGUTH, PJ
MLA Citation
LO, JY, GRISSON, AT, FLOYD, CE, and KORNGUTH, PJ. "COMPUTER-AIDED DIAGNOSIS OF MAMMOGRAMS USING AN ARTIFICIAL NEURAL NETWORK - MERGING OF STANDARDIZED INPUT FEATURES FROM THE ACR LEXICON." MEDICAL IMAGING 1995: IMAGE PROCESSING 2434 (1995): 571-578.
Source
wos-lite
Published In
MEDICAL IMAGING 1995: IMAGE PROCESSING
Volume
2434
Publish Date
1995
Start Page
571
End Page
578
DOI
10.1117/12.208729

Prediction of breast cancer malignancy using an artificial neural network.

BACKGROUND: An artificial neural network (ANN) was developed to predict breast cancer from mammographic findings. This network was evaluated in a retrospective study. METHODS: For a set of patients who were scheduled for biopsy, radiologists interpreted the mammograms and provided data on eight mammographic findings as part of the standard mammographic workup. These findings were encoded as features for an ANN. Results of biopsies were taken as truth in the diagnosis of malignancy. The ANN was trained and evaluated using a jackknife sampling on a set of 260 patient records. Performance of the network was evaluated in terms of sensitivity and specificity over a range of decision thresholds and was expressed as a receiver operating characteristic curve. RESULTS: The ANN performed more accurately than the radiologists (P < 0.08) with a relative sensitivity of 1.0 and specificity of 0.59. CONCLUSIONS: An ANN can be trained to predict malignancy from mammographic findings with a high degree of accuracy.

Authors
Floyd, CE; Lo, JY; Yun, AJ; Sullivan, DC; Kornguth, PJ
MLA Citation
Floyd, CE, Lo, JY, Yun, AJ, Sullivan, DC, and Kornguth, PJ. "Prediction of breast cancer malignancy using an artificial neural network." Cancer 74.11 (December 1, 1994): 2944-2948.
PMID
7954258
Source
pubmed
Published In
Cancer
Volume
74
Issue
11
Publish Date
1994
Start Page
2944
End Page
2948

Bayesian restoration of chest radiographs. Scatter compensation with improved signal-to-noise ratio.

OBJECTIVES: The authors introduce a Bayesian algorithm for digital chest radiography that increases the signal-to-noise ratio, and thus detectability, for low-contrast objects. METHOD: The improved images are formed as a maximum a posteriori probability estimation of a scatter-reduced (contrast-enhanced) image with decreased noise. Noise is constrained by including prior knowledge of image smoothness. Variations between neighboring pixels are penalized for small variations (to suppress Poisson noise), but not for larger variations (to avoid affecting anatomical structure). The technique was optimized to reduce residual scatter in digital radiographs of an anatomical chest phantom. RESULTS: The contrast in the lung was improved by a factor of two, whereas signal-to-noise ratio was improved by a factor of 1.8. Image resolution was unaffected for objects with a contrast greater than 2%. CONCLUSION: This statistical estimation technique shows promise for improving object detectability in radiographs by simultaneously increasing contrast, while constraining noise.

Authors
Floyd, CE; Baydush, AH; Lo, JY; Bowsher, JE; Ravin, CE
MLA Citation
Floyd, CE, Baydush, AH, Lo, JY, Bowsher, JE, and Ravin, CE. "Bayesian restoration of chest radiographs. Scatter compensation with improved signal-to-noise ratio." Invest Radiol 29.10 (October 1994): 904-910.
PMID
7852042
Source
pubmed
Published In
Investigative Radiology
Volume
29
Issue
10
Publish Date
1994
Start Page
904
End Page
910

Scatter compensation in digital chest radiography using the posterior beam stop technique.

A new scatter compensation technique for computed radiography based on posterior beam stop (PBS) sampled scatter measurements and the bicubic spline interpolation technique was proposed. Using only a single exposure, both the clinical image and an array of scatter measurements, which were interpolated into a smooth scatter-only image, were simultaneously acquired. The scatter was subtracted from the clinical image to generate the primary-only image. To gauge the accuracy of scatter estimation, both quantitative and interpolation errors were evaluated. The PBS measurements were compared against the standard beam stop method at 16 locations in an anatomical phantom, resulting in quantitative errors of 2.7% relative to the scatter or 6.8% relative to the primary. Also measured were the interpolation error over 64 interpolation sample locations and 64 midpoint sample locations in the anatomical phantom. The combined interpolation error was 1.9% relative to the scatter or 8.0% relative to the primary. At the interpolation sample locations, the errors were identical between the phantom radiograph and digital portable chest radiographs from five patients. By summing the quantitative and interpolation errors in quadrature, the overall error of the PBS SISTER (scatter interpolation-subtraction technique for radiography) method was 3.3% relative to the scatter or 10% relative to the primary, which was adequate for dual-energy imaging purposes (less than 10% error relative to the scatter or 20% relative to the primary). The change of image contrast, noise, and signal-to-noise ratio (SNR) at six locations in the anatomical phantom were quantitatively analyzed. Contrast and noise were equally enhanced in all anatomical regions, resulting in approximately the same SNR before and after compensation.(ABSTRACT TRUNCATED AT 250 WORDS)

Authors
Lo, JY; Floyd, CE; Baker, JA; Ravin, CE
MLA Citation
Lo, JY, Floyd, CE, Baker, JA, and Ravin, CE. "Scatter compensation in digital chest radiography using the posterior beam stop technique." Med Phys 21.3 (March 1994): 435-443.
PMID
8208219
Source
pubmed
Published In
Medical physics
Volume
21
Issue
3
Publish Date
1994
Start Page
435
End Page
443
DOI
10.1118/1.597388

SPATIALLY VARYING SCATTER COMPENSATION FOR CHEST RADIOGRAPHS USING A HYBRID MADALINE ARTIFICIAL NEURAL-NETWORK

Authors
LO, JY; BAYDUSH, AH; FLOYD, CE
MLA Citation
LO, JY, BAYDUSH, AH, and FLOYD, CE. "SPATIALLY VARYING SCATTER COMPENSATION FOR CHEST RADIOGRAPHS USING A HYBRID MADALINE ARTIFICIAL NEURAL-NETWORK." IMAGE PROCESSING 2167 (1994): 601-611.
Source
wos-lite
Published In
IMAGE PROCESSING
Volume
2167
Publish Date
1994
Start Page
601
End Page
611
DOI
10.1117/12.175095

PREDICTION OF BREAST-CANCER MALIGNANCY FOR DIFFICULT CASES USING AN ARTIFICIAL NEURAL-NETWORK

Authors
FLOYD, CE; YUN, AJ; LO, JY; TOURASSI, G; SULLIVAN, DC; KORNGUTH, PJ; SOC, INTNN
MLA Citation
FLOYD, CE, YUN, AJ, LO, JY, TOURASSI, G, SULLIVAN, DC, KORNGUTH, PJ, and SOC, INTNN. "PREDICTION OF BREAST-CANCER MALIGNANCY FOR DIFFICULT CASES USING AN ARTIFICIAL NEURAL-NETWORK." 1994.
Source
wos-lite
Published In
WORLD CONGRESS ON NEURAL NETWORKS-SAN DIEGO - 1994 INTERNATIONAL NEURAL NETWORK SOCIETY ANNUAL MEETING, VOL 1
Publish Date
1994
Start Page
A127
End Page
A132

Observer evaluation of scatter subtraction for digital portable chest radiographs.

RATIONALE AND OBJECTIVES: The authors compared standard digital portable chest radiographs (DPCXR) to scatter-subtracted DPCXR: METHODS: Thirty DPCXR were obtained using a photostimulable phosphor digital imaging system and a posterior beam stop (PBS) technique that allowed measurement of the scatter component of the DPCXR: The scatter component was subtracted from the clinical image to form a scatter-subtracted image. Six observers recorded preference for the standard image or scatter-subtracted image for identifying five radiographic landmarks and for image quality. RESULTS: A statistically significant preference was demonstrated for the scatter-subtracted images and for viewing the tracheo-bronchial tree, right paratracheal stripe, vertebral column, and support apparatus position. For unprocessed images, there was a statistically significant preference for viewing the pulmonary vasculature. No statistically significant preference was demonstrated for overall image quality. CONCLUSIONS: These results suggest that PBS scatter subtraction holds promise for improving visualization of structures in high-scatter regions of chest radiographs.

Authors
Baker, JA; Floyd, CE; Lo, JY; Ravin, CE
MLA Citation
Baker, JA, Floyd, CE, Lo, JY, and Ravin, CE. "Observer evaluation of scatter subtraction for digital portable chest radiographs." Invest Radiol 28.8 (August 1993): 667-670.
PMID
8375997
Source
pubmed
Published In
Investigative Radiology
Volume
28
Issue
8
Publish Date
1993
Start Page
667
End Page
670

Measurement of scatter fractions in erect posteroanterior and lateral chest radiography.

Scatter fractions (SFs) measured in patients undergoing erect posteroanterior (PA) and lateral chest radiography with a 12:1 antiscatter grid are reported. Modifications to the posterior beam-stop (PBS) technique allowed measurement of scatter in these patients, without altering the diagnostic image and without additional radiation exposure. The SF measurements are reported by anatomic location on 42 clinical chest images. Average SF values ranged from 0.27 to 0.90 on lateral radiographs and from 0.27 to 0.68 on PA radiographs. Scatter measurements with the 12:1 grid were found to be greater than estimates from previous PA chest phantom experiments. To the authors' knowledge, they were the first to measure radiation scatter with the PBS technique in patients undergoing PA and lateral chest radiography with the antiscatter grid.

Authors
Jordan, LK; Floyd, CE; Lo, JY; Ravin, CE
MLA Citation
Jordan, LK, Floyd, CE, Lo, JY, and Ravin, CE. "Measurement of scatter fractions in erect posteroanterior and lateral chest radiography." Radiology 188.1 (July 1993): 215-218.
PMID
8511301
Source
pubmed
Published In
Radiology
Volume
188
Issue
1
Publish Date
1993
Start Page
215
End Page
218
DOI
10.1148/radiology.188.1.8511301

An artificial neural network for estimating scatter exposures in portable chest radiography.

An adaptive linear element (Adaline) was developed to estimate the two-dimensional scatter exposure distribution in digital portable chest radiographs (DPCXR). DPCXRs and quantitative scatter exposure measurements at 64 locations throughout the chest were acquired for ten radiographically normal patients. The Adaline is an artificial neural network which has only a single node and linear thresholding. The Adaline was trained using DPCXR-scatter measurement pairs from five patients. The spatially invariant network would take a portion of the image as its input and estimate the scatter content as output. The trained network was applied to the other five images, and errors were evaluated between estimated and measured scatter values. Performance was compared against a convolution scatter estimation algorithm. The network was evaluated as a function of network size, initial values, and duration of training. Network performance was evaluated qualitatively by the correlation of network weights to physical models, and quantitatively by training and evaluation errors. Using DPCXRs as input, the network learned to describe known scatter exposures accurately (7% error) and estimate scatter in new images (< 8% error) slightly better than convolution methods. Regardless of size and initial shape, all networks adapted into radial exponentials with magnitude of 0.75, perhaps implying an ideal point spread function and average scatter fraction, respectively. To implement scatter compensation, the two-dimensional scatter distribution estimated by the neural network is subtracted from the original DPCXR.

Authors
Lo, JY; Floyd, CE; Baker, JA; Ravin, CE
MLA Citation
Lo, JY, Floyd, CE, Baker, JA, and Ravin, CE. "An artificial neural network for estimating scatter exposures in portable chest radiography." Med Phys 20.4 (July 1993): 965-973.
PMID
8413040
Source
pubmed
Published In
Medical physics
Volume
20
Issue
4
Publish Date
1993
Start Page
965
End Page
973
DOI
10.1118/1.596978

Scatter compensation for digital chest radiography using maximum likelihood expectation maximization.

RATIONALE AND OBJECTIVES: An iterative maximum likelihood expectation maximization algorithm (MLEM) has been developed for scatter compensation in chest radiography. METHODS: The MLEM technique produces a scatter-reduced image which maximizes the probability of observing the measured image. We examined the scatter content and the low-contrast signal-to-noise ratio (SNR) in digital radiographs of anatomical phantoms before and after compensation. RESULTS: MLEM converged to an accurate (6.4% RMS residual scatter error) estimate within 12 iterations. Both contrast and noise were increased in the processed images as iteration progressed. In the lung, contrast was increased 108% and SNR was improved by 10%. In the retrocardiac region, contrast was increased 180% while SNR decreased by 6%. CONCLUSIONS: This is the first report of a post-acquisition scatter compensation technique which can increase SNR. These results suggest that statistical estimation techniques can enhance image quality and quantitative accuracy for digital chest radiography.

Authors
Floyd, CE; Baydush, AH; Lo, JY; Bowsher, JE; Ravin, CE
MLA Citation
Floyd, CE, Baydush, AH, Lo, JY, Bowsher, JE, and Ravin, CE. "Scatter compensation for digital chest radiography using maximum likelihood expectation maximization." Invest Radiol 28.5 (May 1993): 427-433.
PMID
8496036
Source
pubmed
Published In
Investigative Radiology
Volume
28
Issue
5
Publish Date
1993
Start Page
427
End Page
433

SCATTER REDUCTION IN PORTABLE DIGITAL CHEST RADIOGRAPHY WITH BAYESIAN IMAGE ESTIMATION

Authors
BAYDUSH, AH; FLOYD, CE; LO, JY; BOWSHER, JE; RAVIN, CE
MLA Citation
BAYDUSH, AH, FLOYD, CE, LO, JY, BOWSHER, JE, and RAVIN, CE. "SCATTER REDUCTION IN PORTABLE DIGITAL CHEST RADIOGRAPHY WITH BAYESIAN IMAGE ESTIMATION." RADIOLOGY 185 (November 1992): 305-305.
Source
wos-lite
Published In
Radiology
Volume
185
Publish Date
1992
Start Page
305
End Page
305

SPATIALLY VARYING SCATTER ESTIMATION IN PORTABLE CHEST RADIOGRAPHY WITH AN ARTIFICIAL NEURAL NETWORK

Authors
LO, JY; FLOYD, CE; BOWSHER, JE; RAVIN, CE
MLA Citation
LO, JY, FLOYD, CE, BOWSHER, JE, and RAVIN, CE. "SPATIALLY VARYING SCATTER ESTIMATION IN PORTABLE CHEST RADIOGRAPHY WITH AN ARTIFICIAL NEURAL NETWORK." RADIOLOGY 185 (November 1992): 300-300.
Source
wos-lite
Published In
Radiology
Volume
185
Publish Date
1992
Start Page
300
End Page
300

Measurement of scatter fractions in clinical bedside radiography.

The authors present measurements of scatter fraction (SF), the ratio of scattered to total imaged photons, from clinical bedside radiographs of 102 patients. These measurements were obtained by using a new posterior beam-stop technique that does not alter the diagnostic image but that simultaneously provides SF measurements at 224 locations in the image. The SF values in the lung were found to be consistent with previous measurements, while the SF values in the mediastinal and retrocardiac areas were larger than previously reported. SFs in diseased lung were significantly larger than SFs in normal lung. The range of SF values was large for all anatomic locations. For applications in which accurate scatter estimation is required, this wide range of values suggests that SFs should be measured in each individual image.

Authors
Floyd, CE; Baker, JA; Lo, JY; Ravin, CE
MLA Citation
Floyd, CE, Baker, JA, Lo, JY, and Ravin, CE. "Measurement of scatter fractions in clinical bedside radiography." Radiology 183.3 (June 1992): 857-861.
PMID
1584947
Source
pubmed
Published In
Radiology
Volume
183
Issue
3
Publish Date
1992
Start Page
857
End Page
861
DOI
10.1148/radiology.183.3.1584947

Posterior beam-stop method for scatter fraction measurement in digital radiography.

The authors presented a new posterior beam-stop (PBS) technique for measuring the ratio of scattered to total-detected photon flux (scatter fraction) in a radiographic examination while preserving the diagnostic quality of the image. The scatter measurement was made using a standard imaging geometry with both beam stops and an additional x-ray detector placed behind the standard imaging detector. This PBS geometry differs from the standard beam-stop (SBS) technique for scatter measurement. With SBS, a beam-stop shadow appears on the image. To evaluate the PBS technique, scatter fraction measurements were performed on an anatomic phantom using both the PBS and SBS techniques. When compared with the standard technique, PBS provided accurate estimation of scatter fractions. Since the measurement can be performed without degrading a standard clinical radiographic examination, the PBS technique allows simultaneous acquisition of scatter measurements from human patients in combination with a standard radiographic examination.

Authors
Floyd, CE; Baker, JA; Lo, JY; Ravin, CE
MLA Citation
Floyd, CE, Baker, JA, Lo, JY, and Ravin, CE. "Posterior beam-stop method for scatter fraction measurement in digital radiography." Invest Radiol 27.2 (February 1992): 119-123.
PMID
1601602
Source
pubmed
Published In
Investigative Radiology
Volume
27
Issue
2
Publish Date
1992
Start Page
119
End Page
123

Quantitative scatter measurement in digital radiography using a photostimulable phosphor imaging system.

X-ray scatter fractions measured with two detectors are compared: a photostimulable phosphor system (PSP) and a conventional film-screen technique. For both detection methods, a beam-stop technique was used to estimate the scatter fraction in polystyrene phantoms. These scatter fraction measurements are compared to previously reported film-based measurements. Scatter fractions obtained with the PSP were in good agreement both with measurements using film as well as with most previously reported measurements. For the PSP measurements, repeatability was better than 1%. It was found that the PSP provides a precise x-ray detector for quantitative scatter measurement in digital radiography.

Authors
Floyd, CE; Lo, JY; Chotas, HG; Ravin, CE
MLA Citation
Floyd, CE, Lo, JY, Chotas, HG, and Ravin, CE. "Quantitative scatter measurement in digital radiography using a photostimulable phosphor imaging system." Med Phys 18.3 (May 1991): 408-413.
PMID
1870483
Source
pubmed
Published In
Medical physics
Volume
18
Issue
3
Publish Date
1991
Start Page
408
End Page
413
DOI
10.1118/1.596687

ARTIFICIAL NEURAL NETWORKS FOR SPECT IMAGE-RECONSTRUCTION WITH OPTIMIZED WEIGHTED BACKPROJECTION

Authors
FLOYD, CE; BOWSHER, JE; MUNLEY, MT; TOURASSI, GD; GARG, S; BAYDUSH, AH; LO, JY; COLEMAN, RE; IEEE,
MLA Citation
FLOYD, CE, BOWSHER, JE, MUNLEY, MT, TOURASSI, GD, GARG, S, BAYDUSH, AH, LO, JY, COLEMAN, RE, and IEEE, . "ARTIFICIAL NEURAL NETWORKS FOR SPECT IMAGE-RECONSTRUCTION WITH OPTIMIZED WEIGHTED BACKPROJECTION." 1991.
Source
wos-lite
Published In
CONFERENCE RECORD OF THE 1991 IEEE NUCLEAR SCIENCE SYMPOSIUM AND MEDICAL IMAGING CONFERENCE, VOLS 1-3
Publish Date
1991
Start Page
2184
End Page
2188
DOI
10.1109/NSSMIC.1991.259306

Scatter fractions in AMBER imaging.

Images of two phantoms were obtained with use of an advanced multiple-beam equalization radiography system, and scatter fractions were estimated with use of a photostimulable phosphor imaging system. Scatter fractions in the equalized images were lower in the mediastinum-equivalent areas and higher in the lung-equivalent areas, relative to images that were conventionally acquired with use of an antiscatter grid. The differences are attributed to a reduction in incident exposure in the lungs and the presence of cross-scatter between lung and mediastinal regions.

Authors
Chotas, HG; Floyd, CE; Dobbins, JT; Lo, JY; Ravin, CE
MLA Citation
Chotas, HG, Floyd, CE, Dobbins, JT, Lo, JY, and Ravin, CE. "Scatter fractions in AMBER imaging." Radiology 177.3 (December 1990): 879-880.
PMID
2244003
Source
pubmed
Published In
Radiology
Volume
177
Issue
3
Publish Date
1990
Start Page
879
End Page
880
DOI
10.1148/radiology.177.3.2244003
Show More

Research Areas:

  • Breast Diseases
  • Breast Neoplasms
  • Clinical Trials as Topic
  • Computer Simulation
  • Contrast Media
  • Decision Making, Computer-Assisted
  • Decision Support Systems, Clinical
  • Decision Support Techniques
  • Education
  • Evaluation Studies as Topic
  • Image Processing, Computer-Assisted
  • Imaging, Three-Dimensional
  • Machine learning
  • Mammography
  • Models, Psychological
  • Models, Structural
  • Pattern Recognition, Automated
  • ROC Curve
  • Radiographic Image Interpretation, Computer-Assisted
  • Radiology
  • Signal-To-Noise Ratio
  • Technology Assessment, Biomedical
  • Tomosynthesis