Conant, Emily F.

e13538 Background: Breast density is considered a well-established breast cancer risk factor. As quasi-3D, digital breast tomosynthesis (DBT) becomes increasingly utilized for screening, there is an opportunity to routinely estimate volumetric breast density (VBD). However, current methods extrapolate VBD from 2D images acquired with DBT and/or depend on the existence of raw DBT data, which is rarely archived due to cost and storage constraints. Using a racially diverse screening cohort, this study evaluates the potential of deep learning for VBD assessment based solely on 3D reconstructed, “for presentation” DBT images. Methods: We retrospectively analyzed 1,080 negative DBT screening exams obtained between 2011 and 2016 from the Hospital of the University of Pennsylvania (racial makeup, 41.2% White, 54.2% Black, 4.6% Other; mean age ± SD, 57 ± 11 years; mean BMI ± SD, 28.7 ± 7.1 kg/m2), for which both 2D raw and 3D reconstructed DBT images (Selenia Dimensions, Hologic Inc) were available. Corresponding 3D reference-standard tissue segmentations were generated from previously validated software that uses both 3D reconstructed slices and raw 2D DBT data to provide VBD metrics, shown to be strongly correlated with VBD measures from MRI image volumes. We based our deep learning algorithm on the U-Net architecture within the open-source Generally Nuanced Deep Learning Framework (GaNDLF) and created a 3-label image segmentation task (background, dense tissue, and fatty tissue). Our dataset was randomly split into training (70%), validation (15%) and test (15%) sets. We report on the performance of our deep learning algorithm against corresponding reference-standard segmentations for a cranio-caudal (CC) view-only subset. We also stratify our results by the two main racial groups (White and Black). Our evaluation measure was the weighted Dice score (DSC), with 0 signifying no overlap and 1 signifying perfect overlap, overall and separately for each label. Results: Our deep learning algorithm achieved an overall DSC of 0.682 (STD = 0.136). It accurately segmented the three labels of background, fatty tissue, and dense tissue, with DSC scores of 0.995, 0.884, and 0.617, respectively. DSC for White and Black women were 0.688 (STD = 0.127) and 0.680 (STD = 0.146), respectively. Conclusions: Our preliminary analysis suggests that deep learning shows promise in the estimation of VBD using 3D DBT reconstructed, “for presentation” CC view images and does not demonstrate bias among racial groups. Future work involving optimization of performance in other breast views as well as transfer learning based on ground truth masks by clinical radiologists could further enhance this method. In view of rapid clinical conversion to DBT screening, such a tool has the potential to enable large retrospective epidemiological and personalized risk assessment studies of breast density with DBT.

Digital mammography has seen an explosion in the number of radiomic features used for risk-assessment modeling. However, having more features is not necessarily beneficial, as some features may be overly sensitive to imaging physics (contrast, noise, and image sharpness). To measure the effects of imaging physics, we analyzed the feature variation across imaging acquisition settings (kV, mAs) using an anthropomorphic phantom. We also analyzed the intra-woman variation (IWV), a measure of how much a feature varies between breasts with similar parenchymal patterns—a woman’s left and right breasts. From 341 features, we identified “robust” features that minimized the effects of imaging physics and IWV. We also investigated whether robust features offered better case-control classification in an independent data set of 575 images, all with an overall BI-RADS® assessment of 1 (negative) or 2 (benign); 115 images (cases) were of women who developed cancer at least one year after that screening image, matched to 460 controls. We modeled cancer occurrence via logistic regression, using cross-validated area under the receiver-operating-characteristic curve (AUC) to measure model performance. Models using features from the most-robust quartile of features yielded an AUC = 0.59, versus 0.54 for the least-robust, with p < 0.005 for the difference among the quartiles.

Abstract Accurate estimation of mammographic breast density could aid in augmenting breast cancer risk assessment for women undergoing breast screening with full-field digital mammography (FFDM). Breast density can be estimated from FFDM and is most commonly assessed in the clinic by visual grading into one of the four categories defined by the American College of Radiology BI-RADS. However, BI-RADS density assessment is highly subjective and does not provide a quantitative, continuous measure of percent density (PD), which would allow for more refined risk stratification and assessment of density changes. Here, we introduce Deep-LIBRA, an artificial intelligence (AI) tool for fully-automated assessment of breast PD from FFDM images. Two key modules form the core of Deep-LIBRA: 1) an implementation of a modified U-Net architecture for breast segmentation and 2) a radiomic machine learning module that performs PD estimation within the segmented breast region. To develop and validate Deep-LIBRA, raw (i.e., "For Processing") FFDM images (Selenia Dimensions, Hologic Inc.) acquired at two breast cancer screening practices were retrospectively analyzed. For the breast segmentation module, we used a total of 12,100 FFDM studies from 2,200 individual women and a 90%-10% split-sample training-validation approach, using the Dice coefficient to evaluate the accuracy of Deep-LIBRA versus ground-truth manual breast segmentation. For the PD estimation module we used a total of 3,304 FFDM images from 1,652 individual women; manual PD scores obtained with the widely used Cumulus software were used as the "gold standard" in a three-fold cross-validation setting to assess the accuracy of Deep-LIBRA in PD estimation. PD estimates from Deep-LIBRA were also compared with breast density estimates from the commercially available Volpara software. Breast segmentation had a Dice coefficient of 95.31% when compared to ground-truth manual breast segmentation in the validation set. Deep-LIBRA average differences from ground-truth PD scores in the three cross-validation folds were 4.91%, 4.65%, and 4.22%, while Volpara had corresponding average differences of 6.20%, 6.01%, and 5.94%. Deep-LIBRA PD scores were also significantly different from Volpara PD (t-test p-value &lt; 0.001 in all three folds). Preliminary evaluation results show that Deep-LIBRA is a promising AI approach for accurately assessing PD from FFDM images. Citation Format: Omid Haji Maghsoudi, Scott Christopher, Aimilia Gastounioti, Lauren Pantalone, Fang-Fang Wu, Eric A. Cohen, Winham Stacey, Emily F. Conant, Celine Vachon, Despina Kontos. Deep-LIBRA: An artificial intelligence approach for fully-automated assessment of breast density in digital mammography [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2600.