Unifying the Electron Microscopy Multiverse through a Large-scale Foundation Model

Unifying the Electron Microscopy Multiverse through a Large-scale Foundation Model | 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 Article Unifying the Electron Microscopy Multiverse through a Large-scale Foundation Model Lei Ma, Liuyuan He, Ruohua Shi, Wenyao Wang, Cai Yu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6820885/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Accurate analysis of electron microscopy (EM) images is essential for exploring nanoscale biological structures, yet data heterogeneity and fragmented workflows hinder scalable insights. Pretrained on large, diverse datasets, image foundation models provide a robust framework for learning transferable representations across tasks. Here, we introduce EM-DINO, the first image foundational model pretrained on EM-5M, the largest standardized EM corpus (5 million images) encompassing multiple species, tissues, protocols, and resolutions. EM-DINO’s multi-scale embeddings capture rich image features that support multiple applications, including organ-specific pattern recognition, image deduplication, and high quality image restoration. Building on these representations, we developed OmniEM, a U-shaped architecture for unified dense prediction that exceeds task-specific models in both image restoration and segmentation. In restoration benchmarks, OmniEM matches the performance of the EM-specific diffusion model while producing fewer hallucinations that could mislead EM interpretation. It also outperforms previous methods in both generalized mitochondrial segmentation and multi-class organelle segmentation. Furthermore, we demonstrate OmniEM’s integrated capability to generate high-resolution segmentations from low-resolution inputs, offering the potential to enable fine-scale subcellular analysis in legacy and high-throughput EM datasets. Together, EM-5M, EM-DINO, OmniEM, and an integrated Napari plugin comprise a comprehensive end-to-end toolkit for standardized EM analysis, advance cellular and subcellular understanding and accelerating the discovery of novel organelle morphologies and disease-related alterations. Biological sciences/Computational biology and bioinformatics/Software Biological sciences/Computational biology and bioinformatics/Image processing Biological sciences/Biological techniques/Microscopy/Scanning electron microscopy Full Text Additional Declarations There is NO Competing Interest. Cite Share Download PDF Status: Under Review 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-6820885","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":467213397,"identity":"62eedfc8-d12c-43bd-8eed-251d0bbf7153","order_by":0,"name":"Lei Ma","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxklEQVRIiWNgGAWjYBACPgbGBiBlwyAB4vEQo4UNoiWNJC1gcJgULezNDUw3as7nSc5IYHzwto1B3pygFp6DDcw5x24XS0skMBvObWMw3NlASItEIlAL2+3EeRIJbNK8bQwJBgcIaZF/CNTy7xxIC/tv4rRIMDYw57YdSJwNtIWZOC08iQ2Hc/uSiyV7HjZLzjknYbiBkBZ+9uMPH+d8s8uTOJ588MObMht5graAAEhNAgMkTiWIUA8FCcQrHQWjYBSMghEHACEtOka6PdBxAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0001-6024-3854","institution":"Peking Univeristy","correspondingAuthor":true,"prefix":"","firstName":"Lei","middleName":"","lastName":"Ma","suffix":""},{"id":467213398,"identity":"964a8573-c24c-4771-bc9b-9d42d1e38c53","order_by":1,"name":"Liuyuan He","email":"","orcid":"","institution":"Peking Univeristy","correspondingAuthor":false,"prefix":"","firstName":"Liuyuan","middleName":"","lastName":"He","suffix":""},{"id":467213399,"identity":"6504c462-2f57-46f4-8dba-67e8f8ff28ae","order_by":2,"name":"Ruohua Shi","email":"","orcid":"","institution":"Peking Univerisity","correspondingAuthor":false,"prefix":"","firstName":"Ruohua","middleName":"","lastName":"Shi","suffix":""},{"id":467213400,"identity":"af9bb9b9-7232-45c7-a2de-c59c6d0d9d78","order_by":3,"name":"Wenyao Wang","email":"","orcid":"","institution":"Peking University","correspondingAuthor":false,"prefix":"","firstName":"Wenyao","middleName":"","lastName":"Wang","suffix":""},{"id":467213401,"identity":"c069a46c-6693-472f-8d96-459b3ad5d670","order_by":4,"name":"Cai Yu","email":"","orcid":"","institution":"Peking University","correspondingAuthor":false,"prefix":"","firstName":"Cai","middleName":"","lastName":"Yu","suffix":""}],"badges":[],"createdAt":"2025-06-04 13:15:56","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6820885/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6820885/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108183057,"identity":"24c7ae53-15b6-4499-bc49-1fc7e8f6f7b0","added_by":"auto","created_at":"2026-04-30 08:59:46","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4584296,"visible":true,"origin":"","legend":"Article File","description":"","filename":"UnifyingtheElectronMicroscopyMultiversethroughaLargescaleFoundationModelNatureMethods.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6820885/v1_covered_6a3122d2-5b03-4e96-b780-1fa82cb394be.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Unifying the Electron Microscopy Multiverse through a Large-scale Foundation Model","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6820885/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6820885/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Accurate analysis of electron microscopy (EM) images is essential for exploring nanoscale biological structures, yet data heterogeneity and fragmented workflows hinder scalable insights. Pretrained on large, diverse datasets, image foundation models provide a robust framework for learning transferable representations across tasks. Here, we introduce EM-DINO, the first image foundational model pretrained on EM-5M, the largest standardized EM corpus (5 million images) encompassing multiple species, tissues, protocols, and resolutions. EM-DINO’s multi-scale embeddings capture rich image features that support multiple applications, including organ-specific pattern recognition, image deduplication, and high quality image restoration. Building on these representations, we developed OmniEM, a U-shaped architecture for unified dense prediction that exceeds task-specific models in both image restoration and segmentation. In restoration benchmarks, OmniEM matches the performance of the EM-specific diffusion model while producing fewer hallucinations that could mislead EM interpretation. It also outperforms previous methods in both generalized mitochondrial segmentation and multi-class organelle segmentation. Furthermore, we demonstrate OmniEM’s integrated capability to generate high-resolution segmentations from low-resolution inputs, offering the potential to enable fine-scale subcellular analysis in legacy and high-throughput EM datasets. Together, EM-5M, EM-DINO, OmniEM, and an integrated Napari plugin comprise a comprehensive end-to-end toolkit for standardized EM analysis, advance cellular and subcellular understanding and accelerating the discovery of novel organelle morphologies and disease-related alterations.","manuscriptTitle":"Unifying the Electron Microscopy Multiverse through a Large-scale Foundation Model","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-09 01:55:36","doi":"10.21203/rs.3.rs-6820885/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-methods","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"nmeth","sideBox":"Learn more about [Nature Methods](http://www.nature.com/nmeth)","snPcode":"","submissionUrl":"","title":"Nature Methods","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"44e06aa4-8604-4cf5-b87d-4005f5e4a40b","owner":[],"postedDate":"June 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":49608718,"name":"Biological sciences/Computational biology and bioinformatics/Software"},{"id":49608719,"name":"Biological sciences/Computational biology and bioinformatics/Image processing"},{"id":49608720,"name":"Biological sciences/Biological techniques/Microscopy/Scanning electron microscopy"}],"tags":[],"updatedAt":"2026-05-02T18:35:07+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-09 01:55:36","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6820885","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6820885","identity":"rs-6820885","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}