Global-local consistent semi-supervised segmentation of histopathological image with different perturbations

Xi Guan, Qi Zhu, Liang Sun, Junyong Zhao, Daoqiang Zhang, Peng Wan, Wei Shao

Abstract: Highlights•Develop a global–local consistent pathological image segmentation model.•Apply the generative adversarial network to keep the global structure consistency.•Propose a patch correlation term to maintain the local structural similarity.•Test our method on three cohorts for diverse pathological image segmentation tasks.