{"ID":2883487,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.07539","arxiv_id":"2508.07539","title":"Domain Generalization of Pathological Image Segmentation by Patch-Level and WSI-Level Contrastive Learning","abstract":"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.","short_abstract":"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 impractica...","url_abs":"https://arxiv.org/abs/2508.07539","url_pdf":"https://arxiv.org/pdf/2508.07539v1","authors":"[\"Yuki Shigeyasu\",\"Shota Harada\",\"Akihiko Yoshizawa\",\"Kazuhiro Terada\",\"Naoki Nakazima\",\"Mariyo Kurata\",\"Hiroyuki Abe\",\"Tetsuo Ushiku\",\"Ryoma Bise\"]","published":"2025-08-11T01:38:31Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[]","has_code":false}