Fig. 2

AI-based BI-RADS density classification. A A visual display of the range of BI-RADS density classifications for AI models trained with different architectures and training parameters for 50 patients in the testing set. The radiologist interpretation is displayed in the first row. The average breast density rating across all models and radiologist interpretations is displayed in the last row and was used to order the patients from least dense (left) to most dense (right). B The distribution of predicted breast density labels in the testing set differed for experiments with random class sampling (left) compared with equal class sampling (right) at each minibatch. ****P < .001. E. dense = extremely dense; H. dense = heterogeneously dense [30]. [Reprinted with permission from Elsevier (License Number: 5138920035119)]