Pathology foundation models for training-free case retrieval in salivary gland tumors
The study uses content-based retrieval with existing pathology foundation models to diagnose rare salivary gland tumors without needing new training data. It allows pathologists to search institutional archives by querying a region of interest (ROI) on a slide, retrieving morphologically similar cases and their diagnoses. This provides a low-barrier method for case-based differential diagnosis.