A Clinical Benchmark of Foundation Models: Towards Reliable Morphological Subtyping and Cancer Detection on Real-World Barrett’s Esophagus Data

A Clinical Benchmark of Foundation Models: Towards Reliable Morphological Subtyping and Cancer Detection on Real-World Barrett’s Esophagus Data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Clinical Benchmark of Foundation Models: Towards Reliable Morphological Subtyping and Cancer Detection on Real-World Barrett’s Esophagus Data Azar Kazemi, Julia Slotta-Huspenina, Camillo Saueressig, Jingsong Liu, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8066034/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The applicability of emergent histopathology foundation models (Histo-FMs) to real-world diagnostic problems remains unproven. Given the complexity of clinical tasks and the challenges inherent in real-world data, we utilized Histo-FMs to investigate their utility for diagnosing Barrett’s esophagus (BE) and detecting esophageal adenocarcinoma (EAC), a rare malignancy associated with poor patient outcomes. We benchmarked Histo-FMs for these tasks on a real-world cohort representative of routine diagnostics from normal tissue to EAC (N2EAC). The dataset comprised 3,528 hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) from 790 patients (PAXgene-fixed, paraffin-embedded), processed at magnifications ranging from 5× to 40×. A strong multi-rater agreement was achieved between single-scale models for both morphological subtyping and EAC detection. A multi-magnification, multi-backbone aggregation of the five most expert-consistent single-scale models further improved performance (AUROC of 0.907, F1-score of 0.696, accuracy of 0.795, and κ of 0.651 for morphological subtyping; AUROC of 0.909, F1 score of 0.836, accuracy of 0.959, and κ of 0.673 for EAC detection; p<0.05 for most comparisons), indicating robust concordance with expert evaluation. Performance generalized without fixation-specific fine-tuning, underscoring cross-fixation transferability of Histo-FMs. These findings provide the first clinical validation that Histo-FMs can support reliable BE morphological subtyping and EAC detection on real-world data. Pathology Computer Architecture and Engineering Artificial Intelligence and Machine Learning Biomedical Engineering Foundation Model Validation Multiple Instance learning Diagnostic Tasks Weakly-Supervised Learning Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8066034","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":542036233,"identity":"b012dd22-ecff-4e66-946b-b3e310805a8f","order_by":0,"name":"Azar Kazemi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7UlEQVRIiWNgGAWjYJACxgYQyQzECRUgBnMDKVrOgBiMxGoBs9pQ+VgBf//hZx9n1NTJ6bYzsG54OK82mr8dqOVHxTacWiQOHDOeueEYm7HZYQa2G4nbjufOOMzYwNhz5jZuaw42GDM+YONJ3AbRciy3AaiFmbENtxb5w+yfGR/8k6iHaJlzLHc+IS0Gx3iMGTe2GSRAHNZQk7uBkBbDMzzFjDP7Egy3HWZsu5Fw7EDuRqCWg/j8Inf++GbGnm918mbnDx+7+aOmLnfe+cMHH/yowON9BABHx2Ew8wAx6mGgjhTFo2AUjIJRMEIAANLKXq4T9Ms4AAAAAElFTkSuQmCC","orcid":"","institution":"Institute of General and Surgical Pathology, Technical University of Munich, Munich, Germany","correspondingAuthor":true,"prefix":"","firstName":"Azar","middleName":"","lastName":"Kazemi","suffix":""},{"id":542036234,"identity":"732f6782-6321-47f4-babf-56963b99c473","order_by":1,"name":"Julia Slotta-Huspenina","email":"","orcid":"","institution":"Institute of General and Surgical Pathology, Technical University of Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Julia","middleName":"","lastName":"Slotta-Huspenina","suffix":""},{"id":542036235,"identity":"195e3b90-301f-4fde-a0ff-a6bbbb93f337","order_by":2,"name":"Camillo Saueressig","email":"","orcid":"","institution":"TUM School of Computation, Information and Technology, Technical University of Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Camillo","middleName":"","lastName":"Saueressig","suffix":""},{"id":542036236,"identity":"66eb7d59-f5ad-4bd4-81b0-1734125e8cc6","order_by":3,"name":"Jingsong Liu","email":"","orcid":"","institution":"Institute of General and Surgical Pathology, Technical University of Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jingsong","middleName":"","lastName":"Liu","suffix":""},{"id":542036237,"identity":"a2e303ac-47f0-4232-acfe-c983a55aaa63","order_by":4,"name":"Julia Horstmann","email":"","orcid":"","institution":"Klinikum Rechts Der Isar, Medical Clinic and Polyclinic II, Technical University of Munich, Munich, Germany.","correspondingAuthor":false,"prefix":"","firstName":"Julia","middleName":"","lastName":"Horstmann","suffix":""},{"id":542036238,"identity":"03e1afab-2b3c-4886-a52c-a0bd617f0e8a","order_by":5,"name":"Julius Shakhtour","email":"","orcid":"","institution":"Institute of General and Surgical Pathology, Technical University of Munich, Munich, Germany.","correspondingAuthor":false,"prefix":"","firstName":"Julius","middleName":"","lastName":"Shakhtour","suffix":""},{"id":542036239,"identity":"99cf4e95-870f-4282-8d0e-ac96ec8de1a3","order_by":6,"name":"Nassir Navab","email":"","orcid":"","institution":"Computer Aided Medical Procedures, Technical University of Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Nassir","middleName":"","lastName":"Navab","suffix":""},{"id":542036240,"identity":"97685126-e7d9-4e90-98fa-e586d1dc2050","order_by":7,"name":"Saeid Eslami","email":"","orcid":"","institution":"Department of Medical Informatics, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.","correspondingAuthor":false,"prefix":"","firstName":"Saeid","middleName":"","lastName":"Eslami","suffix":""},{"id":542036241,"identity":"65031c01-86a3-4be4-91ca-151f0537fdd7","order_by":8,"name":"Michael Quante","email":"","orcid":"","institution":"Department of Medicine II, University Medical Center Freiburg, Freiburg, Germany.","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Quante","suffix":""},{"id":542036242,"identity":"294a4693-d023-48ca-9a5e-e82d3c32e8fe","order_by":9,"name":"Peter J. Schüffler","email":"","orcid":"","institution":"Institute of General and Surgical Pathology, Technical University of Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"J.","lastName":"Schüffler","suffix":""}],"badges":[],"createdAt":"2025-11-08 20:39:13","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8066034/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8066034/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95657785,"identity":"0ddea0a8-f592-496b-b81c-d8e7fdeff9c1","added_by":"auto","created_at":"2025-11-11 16:21:57","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2482097,"visible":true,"origin":"","legend":"","description":"","filename":"barrettmanuscriptclean11062025preprintSUBMITfinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8066034/v1_covered_5ac19a64-4ce7-4383-8b23-7851013a1a96.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eA Clinical Benchmark of Foundation Models: Towards Reliable Morphological Subtyping and Cancer Detection on Real-World Barrett’s Esophagus Data\u003c/strong\u003e\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Technical University of Munich","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Foundation Model, Validation, Multiple Instance learning, Diagnostic Tasks, Weakly-Supervised Learning","lastPublishedDoi":"10.21203/rs.3.rs-8066034/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8066034/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe applicability of emergent histopathology foundation models (Histo-FMs) to real-world diagnostic problems remains unproven. Given the complexity of clinical tasks and the challenges inherent in real-world data, we utilized Histo-FMs to investigate their utility for diagnosing Barrett’s esophagus (BE) and detecting esophageal adenocarcinoma (EAC), a rare malignancy associated with poor patient outcomes. We benchmarked Histo-FMs for these tasks on a real-world cohort representative of routine diagnostics from normal tissue to EAC (N2EAC). The dataset comprised 3,528 hematoxylin and eosin (H\u0026amp;E)-stained whole-slide images (WSIs) from 790 patients (PAXgene-fixed, paraffin-embedded), processed at magnifications ranging from 5× to 40×. A strong multi-rater agreement was achieved between single-scale models for both morphological subtyping and EAC detection. A multi-magnification, multi-backbone aggregation of the five most expert-consistent single-scale models further improved performance (AUROC of 0.907, F1-score of 0.696, accuracy of 0.795, and κ of 0.651 for morphological subtyping; AUROC of 0.909, F1 score of 0.836, accuracy of 0.959, and κ of 0.673 for EAC detection; p\u0026lt;0.05 for most comparisons), indicating robust concordance with expert evaluation. Performance generalized without fixation-specific fine-tuning, underscoring cross-fixation transferability of Histo-FMs. These findings provide the first clinical validation that Histo-FMs can support reliable BE morphological subtyping and EAC detection on real-world data.\u003c/p\u003e","manuscriptTitle":"A Clinical Benchmark of Foundation Models: Towards Reliable Morphological Subtyping and Cancer Detection on Real-World Barrett’s Esophagus Data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-11 12:24:18","doi":"10.21203/rs.3.rs-8066034/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c65b1bfc-a392-48a6-8bdb-1875f0c4b85b","owner":[],"postedDate":"November 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":57801099,"name":"Pathology"},{"id":57801100,"name":"Computer Architecture and Engineering"},{"id":57801101,"name":"Artificial Intelligence and Machine Learning"},{"id":57801102,"name":"Biomedical Engineering"}],"tags":[],"updatedAt":"2025-11-11T12:24:18+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-11 12:24:18","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8066034","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8066034","identity":"rs-8066034","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}