External Validation of Deep Learning Models for BI-RADS Breast Density Prediction from Ultrasound Images
This work validates that deep learning models for breast density prediction from ultrasound generalize to external data with different racial composition, but highlights the need for optimization in heterogeneously dense breasts.
Three deep learning models (DenseNet121, ViT-B/32, ResNet50) were externally validated for predicting mammographic breast density from ultrasound images on 2,000 exams. DenseNet121 achieved the highest micro-averaged AUROC of 0.885, with best performance in extremely dense breasts (AUROC 0.868-0.899) and lower performance in heterogeneously dense breasts (0.699-0.729).
We externally validated three deep learning models (DenseNet121, ViT-B/32, and ResNet50) for predicting mammographic breast density from breast ultrasound exams on an independent cohort. The external validation set comprised 2,000 ultrasound exams, including 500 cancer cases defined by an initial negative exam (BI-RADS 1 or 2) followed by a cancer diagnosis within 6 months to 10 years, and 1,500 negative controls matched by manufacturer and study year. Performance was measured using patient-level AUROC across four density categories: A (fatty), B (scattered), C (heterogeneous), and D (extremely dense). As a downstream assessment, we also evaluated 10-year risk prediction by incorporating age and AI-derived density into the Tyrer-Cuzick model and comparing performance against a reference model using age and mammography-reported density. All three models performed best in extremely dense breasts (AUROC 0.868-0.899), with strong performance in fatty (0.814-0.838) and scattered density (0.764-0.799), and lower performance in heterogeneously dense breasts (0.699-0.729). DenseNet121 achieved the highest overall performance (micro-averaged AUROC 0.885), and performance across categories was comparable between internal and external testing. For risk modeling, age combined with AI-derived density yielded a lower AUROC than age combined with mammography-reported density (0.541 vs. 0.570; p = 0.23), with no statistically significant difference. These findings indicate that deep learning models generalize well to external data with different racial composition for breast density assessment. While performance is strongest in extremely dense breasts, heterogeneously dense remains more challenging, highlighting the need for targeted optimization.