SRPL-SFDA: Sam-Guided Reliable Pseudo-Labels For Source-Free Domain Adaptation in medical image segmentation

Open Webpage

Xinya Liu, Jianghao Wu, Tao Lu, Shaoting Zhang, Guotai Wang

Published: 2025, Last Modified: 05 Nov 2025Neurocomputing 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Pseudo-label-guided source-free domain adaptation for medical image segmentation models.•Predictions from the source model are used to generate prompts for SAM to obtain refined pseudo labels.•Test-Time Tri-branch Intensity Enhancement (T3IE) to obtain SAM-compatible inputs for medical images.•State-of-the-art performance was achieved on two multi-domain segmentation datasets.