GFPP-MAE: Gradient-Guided Frequency Reconstruction and Position Predictions Advance MAE for 3D CT Image Segmentation

GFPP-MAE: Gradient-Guided Frequency Reconstruction and Position Predictions Advance MAE for 3D CT 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 GFPP-MAE: Gradient-Guided Frequency Reconstruction and Position Predictions Advance MAE for 3D CT Image Segmentation Yuping Peng, Xing Wu, Xing Xiao, Chengliang Wang, Hongqian Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7711032/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 03 Feb, 2026 Read the published version in Multimedia Systems → Version 1 posted 9 You are reading this latest preprint version Abstract 3D computed tomography (CT) image segmentation has been widely studied due to its important role in disease diagnosis and treatment. Since most existing methods rely on expensive manual annotations, self-supervised learning has been introduced to this task. MAE and its variants in 3D medical image analysis have achieved significant performance improvements, but challenges still exist. Firstly, indiscriminately reconstructing voxels leads to learning unimportant areas and redundant information in CT images. Secondly, failing to fully utilize the positional prior information related to fixed human structures in 3D CT images. To address these challenges, this paper proposes the GFPP-MAE model, which consists of the gradient-guided frequency re-construction module (GFRM), the absolute position prediction module (APPM) and the relative position prediction module (RPPM). GFRM reconstructs CT images in the frequency domain and utilizes gradient-guided weighted loss to focus on important edge areas, which helps to avoid learning redundant features and concentrate modeling capability. APPM and RPPM are used to enhance spatial structure perception. APPM learns global structure information by predicting the absolute position. RPPM understands local-global structure consistency by predicting the volume proportion of randomly cropped sub-volume in each base block. The experiments of abdominal multi-organ segmentation on the BTCV dataset and lung tumor segmentation on the MSD Lung dataset both demonstrate that the GFPP-MAE outperforms other state-of-the-art models. The code is available in https://github.com/Dmitvna/GFPP-MAE.git. Self-supervised Learning Masked Autoencoders Image Segmentation 3D CT Images Full Text Additional Declarations Competing interest reported. This work is supported by The Chongqing Technology Innovation & Application Development Key Project (No.CSTB2022TIAD-KPX0167) and Southwest Hospital Cultivation Project of Clinical Innovative Technologies(2025CXJS26). 20 Cite Share Download PDF Status: Published Journal Publication published 03 Feb, 2026 Read the published version in Multimedia Systems → Version 1 posted Editorial decision: Revision requested 12 Nov, 2025 Reviews received at journal 05 Nov, 2025 Reviews received at journal 04 Nov, 2025 Reviewers agreed at journal 30 Oct, 2025 Reviewers agreed at journal 29 Oct, 2025 Reviewers invited by journal 28 Oct, 2025 Editor assigned by journal 27 Oct, 2025 Submission checks completed at journal 27 Sep, 2025 First submitted to journal 25 Sep, 2025 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-7711032","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":540397794,"identity":"b3b42b56-9046-4117-a5bb-85c3e176813e","order_by":0,"name":"Yuping Peng","email":"","orcid":"","institution":"Chongqing University","correspondingAuthor":false,"prefix":"","firstName":"Yuping","middleName":"","lastName":"Peng","suffix":""},{"id":540397795,"identity":"cbbbeeeb-63ea-4ecc-8cd7-66a6e7d8271b","order_by":1,"name":"Xing Wu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwklEQVRIiWNgGAWjYJCCAwwMNhAWDwla0kjUAgSHSdBicCPH8HDBr/PyfDcSGB+8bWOQNyesJS3h8My+24YzbyQwG85tYzDc2UBAi9mN5AOHeXtuJxjcSGCT5m1jSDA4QFBLYgNQyzmQFvbfRGoB2sLz4wDYFmaitNifeZZwmLch2XDmmYfNknPOSRhuIKRFsj3H+DPPHzt5vuPJBz+8KbORJ2gLGDC2AYkDjA1AUoIY9SDwhwGcBkbBKBgFo2AUYAUAaBtFeWP4/IcAAAAASUVORK5CYII=","orcid":"","institution":"Chongqing University","correspondingAuthor":true,"prefix":"","firstName":"Xing","middleName":"","lastName":"Wu","suffix":""},{"id":540397796,"identity":"adba8f99-6ab8-4299-a9ed-148a2fc7e67f","order_by":2,"name":"Xing Xiao","email":"","orcid":"","institution":"Chongqing University","correspondingAuthor":false,"prefix":"","firstName":"Xing","middleName":"","lastName":"Xiao","suffix":""},{"id":540397797,"identity":"1658c52c-cfa9-4c62-857f-5a0459d589f8","order_by":3,"name":"Chengliang Wang","email":"","orcid":"","institution":"Chongqing University","correspondingAuthor":false,"prefix":"","firstName":"Chengliang","middleName":"","lastName":"Wang","suffix":""},{"id":540397798,"identity":"8ce16bdc-45f5-49bb-b3aa-48a32e2640f1","order_by":4,"name":"Hongqian Wang","email":"","orcid":"","institution":"The First Affiliated Hospital of Army Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hongqian","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2025-09-25 09:08:42","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7711032/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7711032/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00530-025-02197-3","type":"published","date":"2026-02-03T15:58:12+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":95283011,"identity":"943e19fa-3b63-4c46-ab2d-2dc600e1a5d6","added_by":"auto","created_at":"2025-11-06 09:33:11","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6961,"visible":true,"origin":"","legend":"","description":"","filename":"1a087cd8a7c04451b02e29de6f2fb071.json","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/e44fd8459a862acace6b793e.json"},{"id":95283010,"identity":"0294a15f-4db0-404f-b257-c4194754aa53","added_by":"auto","created_at":"2025-11-06 09:33:11","extension":"xml","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93019,"visible":true,"origin":"","legend":"","description":"","filename":"1a087cd8a7c04451b02e29de6f2fb0711enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/ceb52fbd8c5074644da3bd1e.xml"},{"id":95283012,"identity":"39c5b01f-ce1e-4324-9d16-b452a1b41245","added_by":"auto","created_at":"2025-11-06 09:33:11","extension":"eps","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":91593,"visible":true,"origin":"","legend":"","description":"","filename":"fig.eps","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/2cfc94d588b60c73fc5e2940.eps"},{"id":95313507,"identity":"2a65b8f5-532c-4ed8-ba93-06eac635e09c","added_by":"auto","created_at":"2025-11-06 15:51:34","extension":"pdf","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2717131,"visible":true,"origin":"","legend":"","description":"","filename":"pengyupingpaper.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/63773bfe32421ab302fcc0c0.pdf"},{"id":95283018,"identity":"823f060a-d59b-4544-92eb-12ac0e943bdf","added_by":"auto","created_at":"2025-11-06 09:33:15","extension":"bst","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35515,"visible":true,"origin":"","legend":"","description":"","filename":"snbasic.bst","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/b4506f9d0bf6c48a13951eed.bst"},{"id":95283016,"identity":"e5fb9f76-b5d1-4e61-af77-644941697b3e","added_by":"auto","created_at":"2025-11-06 09:33:12","extension":"cls","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55857,"visible":true,"origin":"","legend":"","description":"","filename":"snjnl.cls","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/6203d2e9fa6020403ddad06d.cls"},{"id":95283013,"identity":"f56b3816-b1e3-45c6-852d-e2ad74304b02","added_by":"auto","created_at":"2025-11-06 09:33:12","extension":"bst","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64166,"visible":true,"origin":"","legend":"","description":"","filename":"snmathphysnum.bst","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/1514f1f93fb8af29f4dbaf8c.bst"},{"id":95283014,"identity":"f9a59413-399d-49df-ad85-d4694a95d0f0","added_by":"auto","created_at":"2025-11-06 09:33:12","extension":"xml","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":100434,"visible":true,"origin":"","legend":"","description":"","filename":"1a087cd8a7c04451b02e29de6f2fb0711structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/4468657547c9914487ad368f.xml"},{"id":95283019,"identity":"58104f19-0cb4-46ce-a6ea-1c136c62ceca","added_by":"auto","created_at":"2025-11-06 09:33:17","extension":"html","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":110992,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1/8628379d21405efa649cf8da.html"},{"id":102235694,"identity":"c301bafb-b6b3-4498-933b-dbecbb7ee766","added_by":"auto","created_at":"2026-02-09 16:17:30","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1635540,"visible":true,"origin":"","legend":"","description":"","filename":"pengyupingpaper.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7711032/v1_covered_97ff19de-1123-4341-ade1-faa647556217.pdf"}],"financialInterests":"Competing interest reported. This work is supported by The Chongqing Technology Innovation \u0026 Application\nDevelopment Key Project (No.CSTB2022TIAD-KPX0167) and Southwest Hospital\nCultivation Project of Clinical Innovative Technologies(2025CXJS26).\n20","formattedTitle":"GFPP-MAE: Gradient-Guided Frequency Reconstruction and Position Predictions Advance MAE for 3D CT Image Segmentation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"multimedia-systems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mmsj","sideBox":"Learn more about [Multimedia Systems](http://link.springer.com/journal/530)","snPcode":"530","submissionUrl":"https://submission.nature.com/new-submission/530/3","title":"Multimedia Systems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Self-supervised Learning, Masked Autoencoders, Image Segmentation, 3D CT Images","lastPublishedDoi":"10.21203/rs.3.rs-7711032/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7711032/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e3D computed tomography (CT) image segmentation has been widely studied due to its important role in disease diagnosis and treatment. Since most existing methods rely on expensive manual annotations, self-supervised learning has been introduced to this task. MAE and its variants in 3D medical image analysis have achieved significant performance improvements, but challenges still exist. Firstly, indiscriminately reconstructing voxels leads to learning unimportant areas and redundant information in CT images. Secondly, failing to fully utilize the positional prior information related to fixed human structures in 3D CT images. To address these challenges, this paper proposes the GFPP-MAE model, which consists of the gradient-guided frequency re-construction module (GFRM), the absolute position prediction module (APPM) and the relative position prediction module (RPPM). GFRM reconstructs CT images in the frequency domain and utilizes gradient-guided weighted loss to focus on important edge areas, which helps to avoid learning redundant features and concentrate modeling capability. APPM and RPPM are used to enhance spatial structure perception. APPM learns global structure information by predicting the absolute position. RPPM understands local-global structure consistency by predicting the volume proportion of randomly cropped sub-volume in each base block. The experiments of abdominal multi-organ segmentation on the BTCV dataset and lung tumor segmentation on the MSD Lung dataset both demonstrate that the GFPP-MAE outperforms other state-of-the-art models. The code is available in https://github.com/Dmitvna/GFPP-MAE.git.\u003c/p\u003e","manuscriptTitle":"GFPP-MAE: Gradient-Guided Frequency Reconstruction and Position Predictions Advance MAE for 3D CT Image Segmentation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-06 09:33:05","doi":"10.21203/rs.3.rs-7711032/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-13T04:38:20+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-05T12:18:44+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-04T07:48:07+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"18187021491646847408345589910494011096","date":"2025-10-30T08:49:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"221530938335533226382129243112382812489","date":"2025-10-29T10:51:21+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-10-28T09:13:34+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-27T16:51:54+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-27T13:35:26+00:00","index":"","fulltext":""},{"type":"submitted","content":"Multimedia Systems","date":"2025-09-25T09:05:52+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"multimedia-systems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mmsj","sideBox":"Learn more about [Multimedia Systems](http://link.springer.com/journal/530)","snPcode":"530","submissionUrl":"https://submission.nature.com/new-submission/530/3","title":"Multimedia Systems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"fc5c4f3f-cc57-4712-82d3-a31f7054c226","owner":[],"postedDate":"November 6th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-02-09T16:16:41+00:00","versionOfRecord":{"articleIdentity":"rs-7711032","link":"https://doi.org/10.1007/s00530-025-02197-3","journal":{"identity":"multimedia-systems","isVorOnly":false,"title":"Multimedia Systems"},"publishedOn":"2026-02-03 15:58:12","publishedOnDateReadable":"February 3rd, 2026"},"versionCreatedAt":"2025-11-06 09:33:05","video":"","vorDoi":"10.1007/s00530-025-02197-3","vorDoiUrl":"https://doi.org/10.1007/s00530-025-02197-3","workflowStages":[]},"version":"v1","identity":"rs-7711032","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7711032","identity":"rs-7711032","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}