[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-367239-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-367239-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","a-foundation-model-for-generalizable-cancer-diagnosis-and-survival-prediction-from-histopathological-images","A foundation model for generalizable cancer diagnosis and survival prediction from histopathological images","","BEPH is a BEiT-based foundation model for generalizable cancer diagnosis and survival prediction from histopathological images. The model is pre-trained with self-supervised masked image modeling (MIM) on 11 million unlabeled histopathological image patches, learning transferable representations with minimal reliance on expert labels. Those representations are efficiently adapted to patch-level cancer diagnosis, WSI-level cancer classification, and survival prediction across multiple cancer subtypes, aiming to improve performance and broaden clinical deployment of AI.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/a-foundation-model-for-generalizable-cancer-diagnosis-and-survival-prediction-from-histopathological-images/367239/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/a-foundation-model-for-generalizable-cancer-diagnosis-and-survival-prediction-from-histopathological-images/367239.png","ImageObject",300,407,{"name":42,"@type":43},"Levi","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-24","2026-09-23",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",10,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does BEPH target in computational pathology?","Question",{"text":63,"@type":64},"It addresses the limited availability of annotated histopathological data and the morphological differences that reduce the generalization of existing methods across tasks and cancer types.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How is BEPH pre-trained?",{"text":68,"@type":64},"BEPH uses self-supervised masked image modeling (MIM) following a BEiT-based pre-training approach, leveraging 11 million unlabeled histopathological images to learn meaningful representations.",{"name":70,"@type":61,"acceptedAnswer":71},"Which tasks can BEPH representations be adapted to?",{"text":72,"@type":64},"The pre-trained representations are adapted to patch-level cancer diagnosis, WSI-level cancer classification, and survival prediction for multiple cancer subtypes.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},367239,1790220137,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":55,"slug":125},"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":109,"preview_page_count":33,"preview_urls":145},7971461740909,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","Article [https://doi.org/10.1038/s41467-025-57587-y](https://doi.org/10.1038/s41467-025-57587-y)  \nA foundation model for generalizable cancer diagnosis and survival prediction from histopathological images  \nReceived: 1 June 2024  \n\n| Accepted: 21 February 2025 |\n| --- |\n| |\n| Check for updates |\n\nZhaochang Yang 1,8, Ting Wei 1,8, Ying Liang1,2, Xin Yuan 1,3,4,5, RuiTian Gao1, Yujia Xia1, Jie Zhou6, Yue Zhang1,3  & Zhangsheng Yu 1,3,4,6,7   \nComputational pathology, utilizing whole slide images (WSIs) for pathological diagnosis, has advanced the development of intelligent healthcare. However, the scarcity of annotated data and histological differences hinder the general application of existing methods. Extensive histopathological data and the robustness of self-supervised models in small-scale data demonstrate promising prospects for developing foundation pathology models. Here we show BEPH (BEiT-based model Pre-training on Histopathological image), a foundation model that leverages self-supervised learning to learn meaningful representations from 11 million unlabeled histopathological images. These representations are then efﬁciently adapted to various tasks, including patchlevel cancer diagnosis, WSI-level cancer classiﬁcation, and survival prediction for multiple cancer subtypes. By leveraging the masked image modeling (MIM) pre-training approach, BEPH offers an efﬁcient solution to enhance model performance, reduce the reliance on expert annotations, and facilitate the broader application of artiﬁcial intelligence in clinical settings. The pre-trained model is available at [https://github.com/Zhcyoung/BEPH](https://github.com/Zhcyoung/BEPH).  \nHistopathological image analysis has been considered as the gold standard for cancer diagnosis1,2. However, the conventional manual approach to diagnosis is time-consuming and burdensome for pathologists. They need to carefully examine morphological features, such as tissue shape and cell sizes, across the entire slide to provide an accurate diagnosis. Insufﬁcient diagnostic experience can lead to missed diagnoses and misdiagnoses, signiﬁcantly impacting subsequent treatment3.  \nAdvancements in computational pathology and artiﬁcial intelligence offer possibilities for objective diagnosis, prognosis, and treatment response prediction using gigapixel slides4. Particularly, deep learning-based computational pathology has shown promising  \nprospects in various pathological tasks, including disease grading, cancer classiﬁcation, and survival prediction1,5–7. Despite the revolutionary impact of deep learning on pathological images, many challenges still remain to be addressed.  \nThe typical approach for deep learning analysis of pathological images usually involves loading ImageNet pre-trained weights and training the model using WSIs and speciﬁc labels. However, the inherent differences between natural images and pathological images affect the performance8–10. Researchers have recognized the positive impact ofpre-training on pathological images, but their methods often involve supervised pre-training tailored to the speciﬁc tasks. This approach is limited by the scarcity of training data and the challenge of  \n1Department of Bioinformatics and Biostatistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, China. 2Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China. 3SJTU-Yale Joint Center for Biostatistics and Data Science Organization, Shanghai Jiao Tong University, Shanghai, China. 4Center for Biomedical Data Science, Translational Science Institute, Shanghai Jiao Tong University School of Medicine, Shanghai, China. 5National Center for Translational Medicine, Shanghai Jiao Tong University, Shanghai, China. 6School of Mathematical sciences, Shanghai Jiao Tong University, Shanghai, China. 7Clinical Research Institute, Shanghai Jiao Tong University School of Medicine, Shanghai, China. 8These authors co","cbCaip5skYEF4W9o","https://ap.wps.com/l/cbCaip5skYEF4W9o","pdf",9319941,16,"English","# BEPH foundation model and pre-training strategy\n## Self-supervised masked image modeling (MIM)\n## Transfer to patch-level diagnosis and WSI-level classification\n## Survival prediction across cancer subtypes\n## Motivation: annotation scarcity and morphology differences","[{\"question\":\"What problem does BEPH target in computational pathology?\",\"answer\":\"It addresses the limited availability of annotated histopathological data and the morphological differences that reduce the generalization of existing methods across tasks and cancer types.\"},{\"question\":\"How is BEPH pre-trained?\",\"answer\":\"BEPH uses self-supervised masked image modeling (MIM) following a BEiT-based pre-training approach, leveraging 11 million unlabeled histopathological images to learn meaningful representations.\"},{\"question\":\"Which tasks can BEPH representations be adapted to?\",\"answer\":\"The pre-trained representations are adapted to patch-level cancer diagnosis, WSI-level cancer classification, and survival prediction for multiple cancer subtypes.\"}]","A foundation model for generalizable cancer diagnosis and survival prediction from histopathological images | PDF",1790169848,[146,150,154,158],{"page_no":22,"url":147,"width":148,"height":149},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/dd541d645ded616dbb7cbb6f929dedc2.webp",983,1305,{"page_no":26,"url":151,"width":152,"height":153},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/93d717f1850db7f2fee63ed44582df22.webp",982,1304,{"page_no":30,"url":155,"width":156,"height":157},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/3f2dae1fd0bb23f060edf370949dc7ee.webp",981,1303,{"page_no":33,"url":159,"width":160,"height":161},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/ac8053ef6da992c38bf7279fa4864713.webp",980,1301]