PathFL: Multi-alignment Federated Learning for pathology image segmentation

Yuan Zhang

doi:10.1016/j.media.2025.103670

PathFL: Multi-alignment Federated Learning for pathology image segmentation

2025-07-12

0

PRE

AI

DOI：10.1016/j.media.2025.103670

Original

Original request for help

Save

Abstract

En 中文

• We introduce PathFL, a novel multi-alignment federated learning framework for pathology image segmentation. • Image-level collaborative style enhancement enables cross-client style sharing to enhance diversity. • Feature-level adaptive feature alignment ensures consistent representation learning across heterogeneous clients. • Aggregation-level stratified similarity aggregation hierarchically aligns models to enhance global generalization. • PathFL demonstrates superior performance and robust generalizability across cross-source, modality, organ, and scanner scenarios.

PathFL: Multi-alignment Federated Learning for pathology image segmentation

Abstract

Journal

Organization