Jay Baker

Overview:

As a radiologist in the Division of Breast Imaging, I am interested in studying techniques to better detect and assess breast lesions that may represent breast cancer. The major focus of my research activity involves identifying features of breast lesions on mammography/tomosynthesis, ultrasound and MRI that reliably indicate breast cancer or, equally important, reliably indicate a lesion is not breast cancer and biopsy can be safely avoided. 

Breast cancer is the most common malignancy occurring in women and the second most frequent cause of non-skin cancer deaths among women. Screening mammography programs have repeatedly shown a reduction in the mortality from breast cancer by 30 to 50%. However, breast imaging suffers from a lack of specificity. The result is that 60 to 80% of breast biopsies performed in this country are for benign lesions and are therefore - in retrospect - unnecessary. Because of the overlap in imaging features of benign and malignant lesions, however, these lesions cannot be differentiated without tissue sampling, and the extraordinary number of breast biopsies performed markedly increases the cost of breast cancer prevention programs and is an impediment to breast screening for some women. To overcome this limitation, we are working to identify previously unrecognized features of breast lesions.  Some of these features appear to confirm that a lesion is definitively benign without the need for biopsy.  Other features have identified a particular appearance for breast cancer with features that mimic other benign lesions, thus allowing earlier diagnosis and fewer overlooked breast cancers. We are assessing these features with large reader studies to both determine the accuracy and to confirm that the features can be taught and recognized by radiologists at all levels of breast imaging experience.  If successful, widespread adoption and recognition of these features may greatly reduce the number of women who undergo a needle biopsy for a benign breast lesion.  Conversely, widespread recognition of other features may reduce delays in diagnosis of breast cancer.

Positions:

Professor of Radiology

Radiology, Breast Imaging
School of Medicine

Chief, Breast Imaging

Radiology, Breast Imaging
School of Medicine

Member of the Duke Cancer Institute

Duke Cancer Institute
School of Medicine

Education:

B.A. 1988

University of Pennsylvania

M.D. 1992

Duke University

Resident, Radiology

Duke University

Grants:

Breast Elemental Composition Imaging

Administered By
Radiology
Awarded By
National Institutes of Health
Role
Investigator
Start Date
End Date

Publications:

Location, Location, Location

Authors
MLA Citation
Baker, J. A. “Location, Location, Location.” Journal of Breast Imaging, vol. 3, no. 4, July 2021, pp. 405–06. Scopus, doi:10.1093/jbi/wbab047.
URI
https://scholars.duke.edu/individual/pub1499495
Source
scopus
Published In
Journal of Breast Imaging
Volume
3
Published Date
Start Page
405
End Page
406
DOI
10.1093/jbi/wbab047

Impact of Breast Density Notification Legislation on Radiologists' Practices of Reporting Breast Density: A Multi-State Study.

Purpose To evaluate the impact of breast density notification legislation on breast density reporting by radiologists nationally. Materials and Methods The institutional review board exempted this HIPAA-compliant retrospective study from the requirement for informed consent. State-level data over a 5-year period on breast density categorization and breast cancer detection rate were collected from the National Mammography Database (NMD). Z tests were used to calculate differences in proportions. Results Facilities in 13 of 17 states that had breast density notification legislation as of 2014 submitted data to the NMD before and after law enactment. A total of 1 333 541 mammographic studies (hereafter called "mammograms") over a 30-month period, beginning 20 months before and continuing 10 months after law enactment, were included in the analysis. There was a small but statistically significant decrease in the percentage of mammograms reported as showing dense breast tissue (hereafter called "dense mammograms") in the month before law enactment compared with the month after (43.0% [22 338 of 52 000] vs 40.0% [18 604 of 46 464], P < .001). There was no statistically significant difference in the percentage of mammograms reported as dense in the month before law enactment compared with the 10th month after (43.0% [22 338 of 52 000] vs 42.8% [15 835 of 36 991], P = .65). There were no significant differences in the breast cancer detection rate between the month before and the month after law enactment (3.9 vs 3.8 cancers per 1000 mammograms, P = .79) or between the month before law enactment and the 10th month after (3.9 vs 4.2 cancers per 1000 mammograms, P = .55). In 21 analyzed states without breast density notification legislation, the percentage of mammograms reported as dense did not decrease below 42.8% (43 363 of 101 394) from 2010 to 2014, in contrast to 13 analyzed states with breast density notification legislation, which reached a nadir of 39.3% (20 965 of 53 360) (P < .001). Conclusion The percentage of mammograms reported as dense slightly decreased immediately after enactment of breast density notification legislation but then returned to prelegislation percentages within 10 months. (©) RSNA, 2016.
Authors
Bahl, M; Baker, JA; Bhargavan-Chatfield, M; Brandt, EK; Ghate, SV
MLA Citation
Bahl, Manisha, et al. “Impact of Breast Density Notification Legislation on Radiologists' Practices of Reporting Breast Density: A Multi-State Study.Radiology, vol. 280, no. 3, Sept. 2016, pp. 701–06. Pubmed, doi:10.1148/radiol.2016152457.
URI
https://scholars.duke.edu/individual/pub1127450
PMID
27018643
Source
pubmed
Published In
Radiology
Volume
280
Published Date
Start Page
701
End Page
706
DOI
10.1148/radiol.2016152457

Cancelation of MRI guided breast biopsies for suspicious breast lesions identified at 3.0 T MRI: reasons, rates, and outcomes.

RATIONALE AND OBJECTIVES: To determine the cancelation rate of magnetic resonance imaging (MRI)-guided procedures in suspicious breast lesions initially detected at 3.0 Tesla (T) MRI. MATERIALS AND METHODS: With institutional review board approval, a Health Insurance Portability and Accountability Act-compliant retrospective review of 117 suspicious 3.0 T MRI-detected lesions in 101 patients scheduled to undergo MRI-guided procedures was performed; informed consent was waived. Patient information, imaging features, and outcome data were collected and compared among completed and canceled procedures using Fisher's exact test. RESULTS: MRI-guided breast biopsies were canceled in 13% (15/117) because of lesion nonvisualization, including three (20%) masses, one (1%) focus, and 11 (73%) areas of nonmasslike enhancement. Median lesion size was 1.1 cm. Sixty percent (9/15) of nonvisualized lesions were associated with minimal or mild background parenchymal enhancement at MRI. The nonvisualization rate was not associated with patient age, menopausal status, lesion type, size, breast density, or background parenchymal enhancement (P > .7 for each). No cancers were detected at original lesion sites in 14 (93%) patients undergoing follow-up imaging (n = 11) or mastectomy (n = 3) for cancer elsewhere; one (7%) was lost to follow-up. CONCLUSION: The MRI-guided breast biopsy cancelation rate from nonvisualization of suspicious lesions originally detected with 3.0 T MRI scanning was 13%, similar to rates reported for lesions detected at 1.0 and 1.5 T MRI. No cancers were detected on follow-up imaging. Canceling MRI-guided biopsies because of lesion nonvisualization is a reasonable approach if measures are taken to ensure lesion resolution at the time of biopsy and at imaging follow-up.
Authors
MLA Citation
Johnson, Karen S., et al. “Cancelation of MRI guided breast biopsies for suspicious breast lesions identified at 3.0 T MRI: reasons, rates, and outcomes.Acad Radiol, vol. 20, no. 5, May 2013, pp. 569–75. Pubmed, doi:10.1016/j.acra.2013.01.005.
URI
https://scholars.duke.edu/individual/pub932562
PMID
23473719
Source
pubmed
Published In
Acad Radiol
Volume
20
Published Date
Start Page
569
End Page
575
DOI
10.1016/j.acra.2013.01.005

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, Amarpreet S., et al. “Optimized image acquisition for breast tomosynthesis in projection and reconstruction space.Med Phys, vol. 36, no. 11, Nov. 2009, pp. 4859–69. Pubmed, doi:10.1118/1.3231814.
URI
https://scholars.duke.edu/individual/pub719182
PMID
19994493
Source
pubmed
Published In
Medical Physics
Volume
36
Published Date
Start Page
4859
End Page
4869
DOI
10.1118/1.3231814

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, J. L., et al. “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, vol. 6146, June 2006. Scopus, doi:10.1117/12.655235.
URI
https://scholars.duke.edu/individual/pub686572
Source
scopus
Published In
Progress in Biomedical Optics and Imaging Proceedings of Spie
Volume
6146
Published Date
DOI
10.1117/12.655235