BSMAD: Bridging Semantic and Structural Manifolds for Robust Cross-Modality Medical Anomaly Detection

SHIH-CHIH LIN

Submission Type: I want my submission to be considered for both oral and poster presentation.

Keywords: healthcare, zero-shot anomaly detection, vision-language models, medical imaging

TL;DR: BSMAD is a zero-shot vision-language agent bridging CLIP and DINOv3 for robust cross-modality medical anomaly detection

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 439