Enhancing Breast Cancer Detection with Diffusion-Weighted Imaging

A recent study shows that standalone diffusion-weighted imaging (DWI) meaningfully augments AI-based mammography. The study retrospectively analyzed 414 women with pathologically confirmed breast cancer who underwent preoperative mammography and MRI with DWI.

Two radiologists independently reviewed DWI as a standalone adjunct to AI-detected and AI-missed lesions. Notably, when assessed independently, diffusion-weighted imaging identified 83.5% of lesions missed by AI for the primary reader—providing a clear quantitative signal of improved sensitivity when DWI is added to AI-mammography workflows.

Dense breast composition and smaller tumors were the dominant predictors of AI-missed cancers. Specifically, cancers within dense breast tissue and lesions ≤2 cm had higher rates of AI non-detection. Because DWI is sensitive to high cellularity and often shows restricted diffusion in malignancies, it is well suited to reveal these lesions.

Rapid prospective evaluation and harmonization of DWI protocols are needed to confirm generalizability and operational feasibility in screening settings.