The author introduces a Multi-Aspect Vision-Language Matching (MAVL) framework to dissect pathological entities into distinct visual aspects, significantly improving pathology detection.
