Generative feature style augmentation for domain generalization in medical image segmentation

Open Webpage

Yunzhi Huang, Luyi Han, Haoran Dou

Published: 2025, Last Modified: 12 Nov 2025Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•We modeled the domain-level distribution of feature statistics using a generative feature augmentation module for single-source domain generalization in medical image segmentation.•We equipped the model with additional consistency-based constraints to facilitate the network for learning domain-invariant features.•Our proposed model is lightweight and can be seamlessly integrated into existing networks to alleviate performance degradation caused by domain shifts.