In this paper, we address domain shifts in pathological images by focusing on shifts within whole slide images~(WSIs), such as patient characteristics and tissue thickness, rather than shifts between hospitals. Traditional approaches rely on multi-hospital data, but data collection challenges often make this impractical. Therefore, the proposed domain generalization method captures and leverages intra-hospital domain shifts by clustering WSI-level features from non-tumor regions and treating these clusters as domains. To mitigate domain shift, we apply contrastive learning to reduce feature gaps between WSI pairs from different clusters. The proposed method introduces a two-stage contrastive learning approach WSI-level and patch-level contrastive learning to minimize these gaps effectively.
@article{arxiv.2508.07539,
title = {Domain Generalization of Pathological Image Segmentation by Patch-Level and WSI-Level Contrastive Learning},
author = {Yuki Shigeyasu and Shota Harada and Akihiko Yoshizawa and Kazuhiro Terada and Naoki Nakazima and Mariyo Kurata and Hiroyuki Abe and Tetsuo Ushiku and Ryoma Bise},
journal= {arXiv preprint arXiv:2508.07539},
year = {2025}
}