Triad: Vision Foundation Model for 3D Magnetic Resonance Imaging

Triad: Vision Foundation Model for 3D Magnetic Resonance Imaging | 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 Triad: Vision Foundation Model for 3D Magnetic Resonance Imaging Xiaofeng Yang, Shansong Wang, Mojtaba Safari, Qiang Li, Chih-Wei Chang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6129856/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 Vision foundation models (VFMs) are pre-trained on extensive image datasets to learn general representations for diverse types of data. These models can subsequently be fine-tuned for specific downstream tasks, significantly boosting performance across a broad range of applications. However, existing vision foundation models that claim to be applicable to various clinical tasks are mostly pre-trained on 3D computed tomography (CT), which benefits from the availability of extensive 3D CT databases. Significant differences between CT and magnetic resonance imaging (MRI) in imaging principles, signal characteristics, and data distribution may hinder their practical performance and versatility in MRI-specific applications. Here, we propose Triad, a vision foundation model for 3D MRI. Triad adopts a widely used autoencoder architecture to learn robust representations from 131,170 3D MRI volumes and uses organ-independent imaging descriptions to constrain the semantic distribution of the visual modality. The above pre-training dataset is called Triad-131K, which is currently the largest 3D MRI pre-training dataset. We evaluate Triad across three tasks, namely, organ/tumor segmentation, organ/cancer classification, and medical image registration, in two data modalities (within-domain and out-of-domain) settings using 25 downstream datasets. By initializing models with Triad's pre-trained weights, nnUNet-Triad improves segmentation performance by 2.51% compared to nnUNet-Scratch across 17 datasets. Swin-B-Triad achieves a 4.04% improvement over Swin-B-Scratch in classification tasks across five datasets. SwinUNETR-Triad improves by 4.00% compared to SwinUNETR-Scratch in registration tasks across two datasets. Our study demonstrates that pre-training can improve performance when the data modalities and organs of upstream and downstream tasks are consistent. This work highlights the value of large-scale pre-training techniques for downstream tasks in 3D MRI. By open-sourcing Triad's weights, code, and data, we aim to enhance the adaptability and reliability of foundation models for 3D MRI in clinical tasks. Physical sciences/Engineering/Biomedical engineering Biological sciences/Computational biology and bioinformatics/Image processing Biological sciences/Computational biology and bioinformatics/Machine learning 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-6129856","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":423263380,"identity":"e6e616cf-9b2f-480a-82f4-eb535f4c9bad","order_by":0,"name":"Xiaofeng Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABC0lEQVRIiWNgGAWjYDACZiCugPPgLDYCWs7AeXAWPi0oChnbiNBicJz54YMDFXfsGiSSnz38Os8mml/6jAHDh7LDOLVINrMZGxw48yy5QSLN3Fh2W1ruzL4cA8YZ53Br4WdmMJP+2HY4mUEiwUxactvh3A1neAyYedtwa2FjZv8mcfAfSEv6N2nJOf8hWv7i0cLPzGMmcbDhsB2DRI6Z5MeGAxAtjHi0SDbzFBscOHY4gY3nTZk0w7Hk3Jk9bAUHe86l49RicP74xgcHag7b87Onb5P8UWOX28/DvPHBjzJrnFpgILFNIIGBmQfM5jA4QFA9ENgz8B9gYPwBZrM/IEbHKBgFo2AUjBwAABnbVx1k4Ck1AAAAAElFTkSuQmCC","orcid":"","institution":"Emory University","correspondingAuthor":true,"prefix":"","firstName":"Xiaofeng","middleName":"","lastName":"Yang","suffix":""},{"id":423263381,"identity":"f101cf1d-5969-44fb-955d-95e612f383fe","order_by":1,"name":"Shansong Wang","email":"","orcid":"https://orcid.org/0000-0003-4208-2035","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Shansong","middleName":"","lastName":"Wang","suffix":""},{"id":423263382,"identity":"1201210f-ef69-4151-b6fa-317e72a0c5ae","order_by":2,"name":"Mojtaba Safari","email":"","orcid":"https://orcid.org/0000-0003-3295-328X","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"Mojtaba","middleName":"","lastName":"Safari","suffix":""},{"id":423263383,"identity":"22d89354-9bb9-415a-8b4a-9dbf9cbbbf40","order_by":3,"name":"Qiang Li","email":"","orcid":"","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"Qiang","middleName":"","lastName":"Li","suffix":""},{"id":423263384,"identity":"8ca02771-8687-4cf3-ba6b-cd96ceeddd2b","order_by":4,"name":"Chih-Wei Chang","email":"","orcid":"https://orcid.org/0000-0002-3818-4381","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"Chih-Wei","middleName":"","lastName":"Chang","suffix":""},{"id":423263385,"identity":"d1df7289-c566-4dd3-af2c-83b61909a378","order_by":5,"name":"Richard Qiu","email":"","orcid":"","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"Richard","middleName":"","lastName":"Qiu","suffix":""},{"id":423263386,"identity":"538cf8a0-309e-4a35-b95a-c223ba28cc4a","order_by":6,"name":"Justin Roper","email":"","orcid":"","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"Justin","middleName":"","lastName":"Roper","suffix":""},{"id":423263387,"identity":"d5920207-f21a-42e0-bd86-f9bd74a61ae5","order_by":7,"name":"David Yu","email":"","orcid":"","institution":"Emory University","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Yu","suffix":""}],"badges":[],"createdAt":"2025-02-28 15:30:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6129856/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6129856/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":78107392,"identity":"905fa170-3599-4d1e-83c3-21fe2fca39f9","added_by":"auto","created_at":"2025-03-10 04:06:38","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":12284280,"visible":true,"origin":"","legend":"Article File","description":"","filename":"manuscriptV7.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6129856/v1_covered_e6241def-652d-4c4b-8fe1-3eff27aef17c.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Triad: Vision Foundation Model for 3D Magnetic Resonance Imaging","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-6129856/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6129856/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Vision foundation models (VFMs) are pre-trained on extensive image datasets to learn general representations for diverse types of data. These models can subsequently be fine-tuned for specific downstream tasks, significantly boosting performance across a broad range of applications. However, existing vision foundation models that claim to be applicable to various clinical tasks are mostly pre-trained on 3D computed tomography (CT), which benefits from the availability of extensive 3D CT databases. Significant differences between CT and magnetic resonance imaging (MRI) in imaging principles, signal characteristics, and data distribution may hinder their practical performance and versatility in MRI-specific applications. Here, we propose Triad, a vision foundation model for 3D MRI. Triad adopts a widely used autoencoder architecture to learn robust representations from 131,170 3D MRI volumes and uses organ-independent imaging descriptions to constrain the semantic distribution of the visual modality. The above pre-training dataset is called Triad-131K, which is currently the largest 3D MRI pre-training dataset. We evaluate Triad across three tasks, namely, organ/tumor segmentation, organ/cancer classification, and medical image registration, in two data modalities (within-domain and out-of-domain) settings using 25 downstream datasets. By initializing models with Triad's pre-trained weights, nnUNet-Triad improves segmentation performance by 2.51% compared to nnUNet-Scratch across 17 datasets. Swin-B-Triad achieves a 4.04% improvement over Swin-B-Scratch in classification tasks across five datasets. SwinUNETR-Triad improves by 4.00% compared to SwinUNETR-Scratch in registration tasks across two datasets. Our study demonstrates that pre-training can improve performance when the data modalities and organs of upstream and downstream tasks are consistent. This work highlights the value of large-scale pre-training techniques for downstream tasks in 3D MRI. By open-sourcing Triad's weights, code, and data, we aim to enhance the adaptability and reliability of foundation models for 3D MRI in clinical tasks.","manuscriptTitle":"Triad: Vision Foundation Model for 3D Magnetic Resonance Imaging","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-10 03:57:55","doi":"10.21203/rs.3.rs-6129856/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"80d867c7-dc2b-4714-b62b-6ba4c6fa6321","owner":[],"postedDate":"March 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":45105601,"name":"Physical sciences/Engineering/Biomedical engineering"},{"id":45105602,"name":"Biological sciences/Computational biology and bioinformatics/Image processing"},{"id":45105603,"name":"Biological sciences/Computational biology and bioinformatics/Machine learning"}],"tags":[],"updatedAt":"2026-02-07T12:46:34+00:00","versionOfRecord":[],"versionCreatedAt":"2025-03-10 03:57:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6129856","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6129856","identity":"rs-6129856","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}