A Breast Vision Pathology Foundation Model for Real-world Clinical Utility

2026-05-06

AI Summary

Existing foundation models in computational pathology demonstrate strong performance in retrospective studies, yet their efficacy in real-world clinical settings—particularly across the entire breast cancer diagnostic and therapeutic workflow—remains inadequately validated. This work proposes BRAVE, the first pathology vision foundation model specifically designed for breast cancer, trained via self-supervised pretraining on 101,638 whole-slide images from 32 international centers and evaluated across 34 tasks spanning preoperative biopsy, intraoperative frozen section, and postoperative resection. BRAVE delivers the first comprehensive, large-scale, multicenter validation bridging retrospective benchmarks to prospective observational studies, human–AI interaction, and prognostic prediction. Prospective trials show BRAVE safely rules out 76.9% of negative biopsies and 70.1% of negative frozen sections (NPV > 0.95), elevates pathologists’ diagnostic accuracy to 95.1%, and significantly predicts both disease-free and overall survival.

Abstract

Pathology foundation models have shown strong retrospective performance, but whether such systems can support clinically relevant use remains unclear. This challenge is particularly important in breast cancer, where pathological assessment serves as the gold standard for diagnosis and guides treatment planning, surgical decision-making and risk stratification across pre-, intra- and post-operative stages. Here we present \textbf{BRAVE}, a breast-adaptive pathology foundation model developed and evaluated using a total resource of 101,638 breast whole-slide images from 32 sources across Asia, Europe and North America. We assessed BRAVE across 34 tasks in 82 cohorts spanning pre-operative biopsy, intra-operative frozen section and post-operative resection, using an evidence chain comprising retrospective benchmarking, clinically challenging scenarios, workflow-oriented clinical impact simulations, prospective observational validation with the thresholds locked in the retrospective cohorts and crossover pathologist-AI interaction studies. Across these settings, BRAVE supported practical roles in the clinical workflow, including safe exclusion of low-risk cases from routine review, AI-assisted second-review rescue of initially missed positives and prioritization of cases for further assessment. In prospective validation across three centres, BRAVE excluded 76.9% of negative biopsy cases (NPV 0.953) and 70.1% of negative frozen-section cases (NPV 0.973), and triaged 78.8% of post-operative subtyping cases as high-confidence clear-cut cases (NPV 1.000). In reader studies, AI assistance improved balanced accuracy from 88.5% to 95.1% (OR 3.14, P<0.001), with better efficiency, confidence and inter-rater agreement. BRAVE-derived scores also independently predicted disease-free survival (adjusted HR 4.79, P<0.001) and overall survival (adjusted HR 8.14, P<0.001).

Problem

• pathology foundation model
• breast cancer
• clinical utility
• whole-slide image
• real-world validation

Innovation

• foundation model
• breast pathology
• prospective validation
• clinical workflow integration
• AI-assisted diagnosis