Studies have repeatedly shown that breast density, which limits mammographic sensitivity, is also a strong risk factor for breast cancer. As a result, breast density information is increasingly utilized in guiding personalizing breast cancer screening and prevention. Conventional 2D mammography, however, is inherently limited due to the effect of tissue superimposition in estimating breast density. In addition, the commonly used measures of mammographic density cannot fully capture the heterogeneity of the breast parenchymal pattern, shown to be an important additional indicator of breast cancer risk. Breast Tomosynthesis is an emerging 3D x-ray modality which offers superior, tomographic breast tissue visualization compared to 2D mammography. Multiple studies have shown that screening with tomosynthesis reduces recalls while increasing cancer detection compared to screening with mammography alone, which is currently fueling a broad implementation of tomosynthesis for general population breast cancer screening. In addition to improved screening performance, our hypothesis is that 3D measures of breast density and parenchymal pattern complexity from tomosynthesis can outperform density measures from conventional 2D mammography, to improve breast cancer risk estimation. During the previous phase of this award, we developed innovative methods for breast density and parenchymal texture analysis in digital mammography and evaluated their association to breast cancer risk. Our studies have shown compelling evidence that these measures provide powerful new imaging markers that can augment the standard mammographic measures. This renewal application focuses on extending our work to the emerging technology of tomosynthesis, by developing novel measures of breast tissue composition from tomosynthesis and by determining their association to breast cancer risk. Towards this end, we will perform the largest association study reported to-date for tomosynthesis and breast cancer risk, including a well-characterized and diverse sample of 675 cases and 2700 controls nested within two large academic breast cancer screening practices at the University of Pennsylvania (UPenn) and the Mayo Clinic.
In AIM1 we will extend and optimize our mammographic parenchymal analysis software for breast tomosynthesis using sophisticated computational methods pioneered from our group for breast tomosynthesis;
in AIM2 we will determine associations between breast cancer and the novel tomosynthesis density and texture measures, using retrospective data analysis as our training set; and in AIM3 we will perform independent validation using prospectively collected, ethnically diverse, samples from both institutions. Within this setting, we will also evaluate the performance of measures derived from synthetic digital mammograms, increasingly replacing conventional digital mammography images in tomosynthesis acquisition. This study will be the first to evaluate tomosynthesis in cancer risk assessment and will develop the necessary tools to enable larger multi-center studies. Ultimately, these novel biomarkers could lead to more accurate risk prediction, improving personalized breast cancer screening and prevention.

Public Health Relevance

The improved clinical performance achieved with digital breast tomosynthesis is currently fueling a broad adaptation of tomosynthesis for general population breast cancer screening. Our study will be the first to evaluate tomosynthesis imaging measures in breast cancer risk assessment and will develop the necessary technology to enable larger multi-center studies. Ultimately, these novel imaging biomarkers may lead to more accurate individualized risk stratification, improving personalized breast cancer screening and prevention.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1)
Program Officer
Baker, Houston
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Pennsylvania
Schools of Medicine
United States
Zip Code
Gastounioti, Aimilia; Oustimov, Andrew; Hsieh, Meng-Kang et al. (2018) Using Convolutional Neural Networks for Enhanced Capture of Breast Parenchymal Complexity Patterns Associated with Breast Cancer Risk. Acad Radiol 25:977-984
Conant, Emily F; Keller, Brad M; Pantalone, Lauren et al. (2017) Agreement between Breast Percentage Density Estimations from Standard-Dose versus Synthetic Digital Mammograms: Results from a Large Screening Cohort Using Automated Measures. Radiology 283:673-680
Ray, Shonket; Chen, Lin; Keller, Brad M et al. (2016) Association between Breast Parenchymal Complexity and False-Positive Recall From Digital Mammography Versus Breast Tomosynthesis: Preliminary Investigation in the ACRIN PA 4006 Trial. Acad Radiol 23:977-86
Gastounioti, Aimilia; Oustimov, Andrew; Keller, Brad M et al. (2016) Breast parenchymal patterns in processed versus raw digital mammograms: A large population study toward assessing differences in quantitative measures across image representations. Med Phys 43:5862
Chen, Lin; Ray, Shonket; Keller, Brad M et al. (2016) The Impact of Acquisition Dose on Quantitative Breast Density Estimation with Digital Mammography: Results from ACRIN PA 4006. Radiology 280:693-700
Gastounioti, Aimilia; Conant, Emily F; Kontos, Despina (2016) Beyond breast density: a review on the advancing role of parenchymal texture analysis in breast cancer risk assessment. Breast Cancer Res 18:91
McCarthy, Anne Marie; Keller, Brad M; Pantalone, Lauren M et al. (2016) Racial Differences in Quantitative Measures of Area and Volumetric Breast Density. J Natl Cancer Inst 108:
Pertuz, Said; McDonald, Elizabeth S; Weinstein, Susan P et al. (2016) Fully Automated Quantitative Estimation of Volumetric Breast Density from Digital Breast Tomosynthesis Images: Preliminary Results and Comparison with Digital Mammography and MR Imaging. Radiology 279:65-74
McCarthy, Anne Marie; Keller, Brad; Kontos, Despina et al. (2015) The use of the Gail model, body mass index and SNPs to predict breast cancer among women with abnormal (BI-RADS 4) mammograms. Breast Cancer Res 17:1
Zheng, Yuanjie; Keller, Brad M; Ray, Shonket et al. (2015) Parenchymal texture analysis in digital mammography: A fully automated pipeline for breast cancer risk assessment. Med Phys 42:4149-60

Showing the most recent 10 out of 19 publications