Segment Any Tissue: One-shot reference guided training-free automatic point prompting for medical image segmentation

Xueyu Liu, Guangze Shi, Rui Wang, Yexin Lai, Jianan Zhang, Weixia Han, Min Lei, Ming Li, Xiaoshuang Zhou, Yongfei Wu, Chen Wang, Wen Zheng

Abstract: Highlights•SAT segments tissues from medical images using one-shot reference, requiring no training.•A cyclic feedback framework with novel prompts optimizes SAM for medical segmentation.•SAT is validated on six datasets, covering tissues from macro to micro levels.