Med-Banana: Learning Quality-Controlled Medical Image Editing from Success-and-Failure Trajectories 事件

Med-Banana: Learning Quality-Controlled Medical Image Editing from Success-and-Failure Trajectories arXiv:2511.00801v4 Announce Type: replace Abstract: Text-guided medical image editing must satisfy the requested pathology while preserving anatomy, modality-specific appearance, and clinical plausibility. However, existing datasets largely supervise editors with final accepted edits and discard the failed attempts produced during generation. We argue that these failures provide essential supervi