Revealing Mammographic Phenotypes in Deep Learning Breast Cancer Risk Models
For researchers and clinicians using AI in breast cancer risk prediction, this work provides a method to uncover latent imaging phenotypes and confounders, though it is an incremental application of existing clustering techniques to a specific model.
The authors used clustering of patch embeddings from a pre-trained deep learning model (Mirai) to identify recurring mammographic phenotypes associated with 5-year breast cancer risk, revealing both clinically meaningful patterns (dense tissue, microcalcifications) and shortcut artifacts (clips).
Mammogram-based deep learning models have improved breast cancer risk prediction, but the learned imaging patterns remain underexplored. Existing interpretability methods rely on single-image saliency maps, failing to identify recurring mammographic phenotypes across large patient cohorts. By clustering patch embeddings from a pre-trained model, Mirai, we isolate recurring phenotypes linked to 5-year cancer risk. Analyses show risk-increasing phenotypes capture complex structures (e.g., dense tissue, microcalcifications) and shortcut artifacts (e.g., clips). These phenotypes correlate strongly with older age and higher BI-RADS density. Our framework connects tissue patterns to AI risk scores, revealing clinical signatures and potential latent model confounders.