Semi-supervised medical image segmentation via weak-to-strong perturbation consistency and edge-aware contrastive representation

Yang Yang, Guoying Sun, Tong Zhang, Ruixuan Wang, Jingyong Su

Abstract: Highlights•Propose a new semi-supervised learning segmentation framework.•Propose two novel strategies for effectively using unlabeled data.•Achieve improved object edge segmentation performance.•Set new state-of-the-art performance on multiple medical imaging datasets.