Stain-adaptive self-supervised learning for histopathology image analysis

Haili Ye, Yuan-yuan Yang, Shunzhi Zhu, Da-Han Wang, Xu-Yao Zhang, Xin Yang, Heguang Huang

Abstract: Highlights•We propose SASSL: a novel stain-adaptive self-supervised learning method.•SASSL integrates stain domain adversarial training into the SSL framework.•SASSL adapts to stain variations while learning invariant tissue features.•Experiments show SASSL improves histopathology analysis across multiple tasks.