{"id":1186642,"date":"2026-09-17T11:09:22","date_gmt":"2026-09-17T18:09:22","guid":{"rendered":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/publication\/prism2-unlocking-multi-modal-general-pathology-ai-with-clinical-dialogue\/"},"modified":"2026-09-23T12:19:49","modified_gmt":"2026-09-23T19:19:49","slug":"prism2-unlocking-multi-modal-general-pathology-ai-with-clinical-dialogue","status":"publish","type":"msr-research-item","link":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/publication\/prism2-unlocking-multi-modal-general-pathology-ai-with-clinical-dialogue\/","title":{"rendered":"PRISM2: Unlocking Multi-Modal General Pathology AI with Clinical Dialogue"},"content":{"rendered":"\n\n\n<p class=\"wp-block-paragraph\">Recent rapid progress in the field of computational pathology has been enabled by foundation models. These models are beginning to move beyond encoding image patches towards whole-slide understanding but their clinical utility remains limited. In this work, we present PRISM2, a multimodal slide-level foundation model trained on data from 700,000 diagnostic specimen-report pairs, the largest vision (2.3 million whole slide images) and language (14M question-answer pairs) histopathology dataset to date. By learning through clinical-dialogue supervision, PRISM2 aligns histomorphologic features with the language of diagnostic reasoning, producing slide-level representations that support both direct diagnostic question-answering and transferable embeddings for downstream tasks. Without additional training, PRISM2 matches or exceeds the cancer-detection performance of clinical-grade products. This is observed without loss of generality on other tasks, where PRISM2 achieves top performance. Finally, using survival prediction as the example, we show that task-specific finetuning with a large dataset can outperform task-specific models, further improving performance. These results demonstrate how language-supervised pretraining provides a scalable, clinically grounded signal for learning generalizable pathology representations, bridging human diagnostic reasoning and foundation-model performance.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Recent rapid progress in the field of computational pathology has been enabled by foundation models. These models are beginning to move beyond encoding image patches towards whole-slide understanding but their clinical utility remains limited. In this work, we present PRISM2, a multimodal slide-level foundation model trained on data from 700,000 diagnostic specimen-report pairs, the largest [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"text","value":"George Shaikovski","user_id":0},{"type":"text","value":"E. Vorontsov","user_id":0},{"type":"text","value":"Adam Casson","user_id":0},{"type":"text","value":"Julian Viret","user_id":0},{"type":"text","value":"Eric Zimmermann","user_id":0},{"type":"user_nicename","value":"Neil Tenenholtz","user_id":"38464"},{"type":"user_nicename","value":"Yun Wang","user_id":"37827"},{"type":"text","value":"Jan H Bernhard","user_id":0},{"type":"text","value":"R. Godrich","user_id":0},{"type":"text","value":"J. Retamero","user_id":0},{"type":"text","value":"Razik Yousfi","user_id":0},{"type":"user_nicename","value":"Nicolo Fusi","user_id":"31829"},{"type":"text","value":"Thomas J Fuchs","user_id":0},{"type":"user_nicename","value":"Kristen Severson","user_id":"40372"},{"type":"text","value":"Siqi Liu","user_id":0}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"arXiv","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"10.48550\/arXiv.2506.13063","msr_arxiv_id":"2506.13063","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2025-10-31","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2025,"msr_month":10,"msr_day":31,"msr_microsoftintellectualproperty":false,"msr_pub_id":"3e89b4de3f8ecf58bf07367adb14ed5a1316d5b8","msr_publication_uploader":[{"type":"url","title":"https:\/\/dblp.org\/rec\/journals\/corr\/abs-2506-13063.html","label_id":243109,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/arxiv.org\/abs\/2506.13063","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"3e89b4de3f8ecf58bf07367adb14ed5a1316d5b8","msr_s2_pdf_url":"","msr_citation_count_updated":"","msr_citation_count":12,"msr_influential_citations":1,"msr_reference_count":54,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"3e89b4de3f8ecf58bf07367adb14ed5a1316d5b8"},{"provider":"doi","id":"10.48550\/arXiv.2506.13063"},{"provider":"arxiv","id":"2506.13063"},{"provider":"dblp","id":"journals\/corr\/abs-2506-13063"},{"provider":"corpusid","id":"279402003"}],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13562],"msr-publication-type":[270373],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[246691,263185],"msr-conference":[],"msr-journal":[268188],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1186642","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-computer-vision","msr-locale-en_us","msr-field-of-study-computer-science","msr-field-of-study-computer-vision-and-pattern-recognition"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2025-10-31","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"arXiv","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"10.48550\/arXiv.2506.13063","msr_publication_uploader":[{"type":"url","title":"https:\/\/dblp.org\/rec\/journals\/corr\/abs-2506-13063.html","label_id":243109,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/arxiv.org\/abs\/2506.13063","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_citation_count":12,"msr_citation_count_updated":"","msr_s2_paper_id":"3e89b4de3f8ecf58bf07367adb14ed5a1316d5b8","msr_influential_citations":1,"msr_reference_count":54,"msr_arxiv_id":"2506.13063","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"George Shaikovski","user_id":0,"rest_url":false},{"type":"text","value":"E. Vorontsov","user_id":0,"rest_url":false},{"type":"text","value":"Adam Casson","user_id":0,"rest_url":false},{"type":"text","value":"Julian Viret","user_id":0,"rest_url":false},{"type":"text","value":"Eric Zimmermann","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Neil Tenenholtz","user_id":38464,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Neil Tenenholtz"},{"type":"user_nicename","value":"Yun Wang","user_id":37827,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Yun Wang"},{"type":"text","value":"Jan H Bernhard","user_id":0,"rest_url":false},{"type":"text","value":"R. Godrich","user_id":0,"rest_url":false},{"type":"text","value":"J. Retamero","user_id":0,"rest_url":false},{"type":"text","value":"Razik Yousfi","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Nicolo Fusi","user_id":31829,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Nicolo Fusi"},{"type":"text","value":"Thomas J Fuchs","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Kristen Severson","user_id":40372,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Kristen Severson"},{"type":"text","value":"Siqi Liu","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[199563],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"misc","related_content":[],"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1186642","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1186642\/revisions"}],"predecessor-version":[{"id":1187091,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1186642\/revisions\/1187091"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1186642"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1186642"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1186642"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1186642"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1186642"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1186642"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1186642"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1186642"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1186642"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1186642"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1186642"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1186642"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1186642"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1186642"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}