Data-efficient federated semi-supervised learning framework via pseudo supervision refinement strategy for lung tumor segmentation

Weixing Li, Xipeng Pan, Zhen Zhang, Guangyao Wu, Guanchao Ye, Leonard Wee, Andre Dekker, Chu Han, Lei Shi, Zaiyi Liu, Zhenbing Liu, Zhenwei Shi

Abstract: Highlights•Proposes a realistic federated scenario with partially and fully unlabeled sites.•Introduces federated self-supervised pre-training and semi-supervised fine-tuning.•Pseudo Supervision Refinement reduces label noise and stabilizes training process.•Designs Dynamic Model Aggregation to adaptively generate aggregation weights.