PKDG: Prior Knowledge based Domain Generalization Model for fundus image segmentation

PKDG: Prior Knowledge based Domain Generalization Model for fundus image segmentation | 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 PKDG: Prior Knowledge based Domain Generalization Model for fundus image segmentation Jianan Xia, Lili Zeng, Tengfei Li, Jingyan Xue, Chuanchuan Zhao, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4703557/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 3 You are reading this latest preprint version Abstract Convolutional neural network performance is prone to be significantly affected by the domain difference between the training data (source domain) and the test data (target domain).To address this issue and simultaneously to meet the multi-institutional healthcare data privacy protection needs, we propose a Prior Knowledge based Domain Generalization (PKDG) model for segmenting the optic cup and disc regions. Unlike existing methods, we incorporate clinical a priori knowledge into two phases. In the model training phase, a Priori Knowledge Module (PKM) is introduced to fuse client-side high-level information with server-side shared feature to generate a new dataset that simulates the differences between different domains. Further, A priori knowledge constraints are imposed on the model training process to enhance its generalization capability. In the model generalization phase, the Visualization Module (VM) reconstructs the shape of the segmentation results based on the intermediate representations, thus improving the visual interpretability. The proposed PKDG model provides a framework for optic cup and disc segmentation that enables privacy-preserving multicenter collaborative disease supporting diagnosis while ensuring robustness and superior performance in terms of accuracy and visual interpretability. Prior Knowledge Domain generalization Fundus image segmentation Federated learning Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editor assigned by journal 08 Jul, 2024 Submission checks completed at journal 08 Jul, 2024 First submitted to journal 08 Jul, 2024 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-4703557","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":324027578,"identity":"03c25314-fd3a-4add-9c0b-2736948ee280","order_by":0,"name":"Jianan Xia","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Jianan","middleName":"","lastName":"Xia","suffix":""},{"id":324027579,"identity":"35c01e19-3ada-4292-8652-978c58ecb330","order_by":1,"name":"Lili Zeng","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Lili","middleName":"","lastName":"Zeng","suffix":""},{"id":324027580,"identity":"5b33f754-7e9f-422c-a949-bbd8996807a6","order_by":2,"name":"Tengfei Li","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Tengfei","middleName":"","lastName":"Li","suffix":""},{"id":324027581,"identity":"84996952-a789-4181-baa1-81a094f63cda","order_by":3,"name":"Jingyan Xue","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Jingyan","middleName":"","lastName":"Xue","suffix":""},{"id":324027583,"identity":"f0f07c03-49e6-49d0-b607-aee09f8f2e39","order_by":4,"name":"Chuanchuan Zhao","email":"","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Chuanchuan","middleName":"","lastName":"Zhao","suffix":""},{"id":324027584,"identity":"061631e6-aed7-4f60-8035-3f0147662557","order_by":5,"name":"Xuezhong Zhou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYHACNiC2SWBsgLGJ1JKWwNhGopbDCRDVxGgxOH/82YOPbefzmOf3GDB8KDvMwD+7Ab8WyYYD6YYz224XM7bxGDDOOHeYQeLOAfxa+Bkbjknztt1ObARqYeZtO8xgIJFAwCPMjG1ALecgWv4So4WfjZkNqOUARAsjMVoke9jYJGecSwb6Ja3gYM+5dB6JGwS0gEJM4kOZXZ5h8+GND36UWcvxzyCgBQ4MGxgYDgBpHiLVA4E88UpHwSgYBaNgpAEAKw8+e1GQTcYAAAAASUVORK5CYII=","orcid":"","institution":"Beijing Jiaotong University","correspondingAuthor":true,"prefix":"","firstName":"Xuezhong","middleName":"","lastName":"Zhou","suffix":""}],"badges":[],"createdAt":"2024-07-08 07:42:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4703557/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4703557/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":61458811,"identity":"3163f3fd-9cc6-4040-8e81-ca44a950776a","added_by":"auto","created_at":"2024-07-31 04:17:56","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3449018,"visible":true,"origin":"","legend":"","description":"","filename":"PKDGTVC.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4703557/v1_covered_55880a24-f827-4312-bf79-c1ebb3d8b2de.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"PKDG: Prior Knowledge based Domain Generalization Model for fundus image segmentation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"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":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Prior Knowledge, Domain generalization, Fundus image segmentation, Federated learning","lastPublishedDoi":"10.21203/rs.3.rs-4703557/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4703557/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Convolutional neural network performance is prone to be significantly affected by the domain difference between the training data (source domain) and the test data (target domain).To address this issue and simultaneously to meet the multi-institutional healthcare data privacy protection needs, we propose a Prior Knowledge based Domain Generalization (PKDG) model for segmenting the optic cup and disc regions. Unlike existing methods, we incorporate clinical a priori knowledge into two phases. In the model training phase, a Priori Knowledge Module (PKM) is introduced to fuse client-side high-level information with server-side shared feature to generate a new dataset that simulates the differences between different domains. Further, A priori knowledge constraints are imposed on the model training process to enhance its generalization capability. In the model generalization phase, the Visualization Module (VM) reconstructs the shape of the segmentation results based on the intermediate representations, thus improving the visual interpretability. The proposed PKDG model provides a framework for optic cup and disc segmentation that enables privacy-preserving multicenter collaborative disease supporting diagnosis while ensuring robustness and superior performance in terms of accuracy and visual interpretability.","manuscriptTitle":"PKDG: Prior Knowledge based Domain Generalization Model for fundus image segmentation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-07-31 04:09:43","doi":"10.21203/rs.3.rs-4703557/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorAssigned","content":"","date":"2024-07-08T08:44:54+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-07-08T07:49:34+00:00","index":"","fulltext":""},{"type":"submitted","content":"The Visual Computer","date":"2024-07-08T07:41:12+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"3e392e5c-5ca0-4ada-8334-0efca9e81f9f","owner":[],"postedDate":"July 31st, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2024-07-31T04:09:43+00:00","versionOfRecord":[],"versionCreatedAt":"2024-07-31 04:09:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4703557","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4703557","identity":"rs-4703557","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}