Fully automated segmentation of [18F]FDG- and PSMA-PET/CT images via data-centric Deep-Learning. | 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 Fully automated segmentation of [ 18 F]FDG- and PSMA-PET/CT images via data-centric Deep-Learning. Manuel Pires, Sebastian Gutschmayer, Enrica Bergalla, Stephane Chauvie, and 40 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9407530/v2 This work is licensed under a CC BY 4.0 License Status: Posted Version 2 posted You are reading this latest preprint version Show more versions Abstract Purpose To develop and validate LION (Lesion Identification in Oncological Nuclear imaging), an open-source PET-only tumor segmentation pipeline for [ 18 F]FDG and PSMA-targeted PET/CT, and to investigate how training data characteristics influence segmentation performance. Materials and Methods In this retrospective multicenter study, 5,209 [ 18 F]FDG PET/CT scans spanning 19 disease types and 2,046 PSMA-targeted PET/CT scans were used to train PET-only segmentation models. Tumor segmentation incorporated organs with physiological uptake as auxiliary classes to enable PET-only inference. Tumor Occurrence Maps (TOMs) quantified tumor spatial diversity across the training data. For [ 18 F]FDG, disease-specific and mixed-disease models trained on progressively larger subsets were compared to test whether increasing spatial diversity improves generalization. Scanner-related domain shift was analyzed using DINOv2 embeddings. Models were evaluated on multicenter holdout cohorts (616 [ 18 F]FDG across 4 diseases; 443 PSMA-targeted prostate cancer scans) and compared with three open-source tools. Results Organ context improved median Dice from 0.62 to 0.71 for [ 18 F]FDG and from 0.75 to 0.83 for PSMA. Spatial diversity measured by TOMs was strongly associated with Dice (Spearman ρ = 0.80, P = 0.003). A mixed-disease model trained on 500 patients matched the performance of a lymphoma specialist model trained on 3,031 cases. DINOv2 embeddings revealed scanner-induced domain shift, between same-disease cohorts. LION achieved median Dice scores of 0.71 ([ 18 F]FDG) and 0.85 (PSMA) and outperformed other open-source approaches on common holdout patients. Conclusion LION enables PET-only automated segmentation for [ 18 F]FDG and PSMA-targeted PET. Training data composition, particularly spatial diversity quantified by TOMs, was strongly associated with segmentation performance. Nuclear Medicine & Medical Imaging Artificial Intelligence and Machine Learning Full Text Additional Declarations The authors declare potential competing interests as follows: Lalith Kumar Shiyam Sundar and Thomas Beyer are cofounders of Zenta GmbH/Austria. D. Kersting reports a research grant from Pfizer, personal fees from Alnylam, GE Healthcare, Novartis, and Pfizer, and travel grants from Life Molecular Imaging and Sanofi, all outside of the submitted work. Rudolf A. Werner has received speaker honoraria from Novartis. Supplementary Files SupplementaryMaterials.docx Cite Share Download PDF Status: Posted Version 2 posted You are reading this latest preprint version Show more versions 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-9407530","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":623395652,"identity":"f3e6973b-1b8c-4f60-87e6-c1a30e31ede5","order_by":0,"name":"Manuel Pires","email":"data:image/png;base64,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","orcid":"","institution":"QIMP Team, Medical University of Vienna, Vienna, Austria","correspondingAuthor":true,"prefix":"","firstName":"Manuel","middleName":"","lastName":"Pires","suffix":""},{"id":623395653,"identity":"58fa4136-d431-49a6-8af6-58eb0f97551e","order_by":1,"name":"Sebastian Gutschmayer","email":"","orcid":"","institution":"QIMP Team, Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Sebastian","middleName":"","lastName":"Gutschmayer","suffix":""},{"id":623395654,"identity":"e32e2406-4047-485b-bad6-ab5347e1218f","order_by":2,"name":"Enrica Bergalla","email":"","orcid":"","institution":"Santa Croce e Carle Hospital, Medical Physics, Cuneo, Italy","correspondingAuthor":false,"prefix":"","firstName":"Enrica","middleName":"","lastName":"Bergalla","suffix":""},{"id":623395655,"identity":"fc183091-713a-4b7f-a70d-3503ce00545c","order_by":3,"name":"Stephane Chauvie","email":"","orcid":"","institution":"Santa Croce e Carle Hospital, Medical Physics, Cuneo, Italy","correspondingAuthor":false,"prefix":"","firstName":"Stephane","middleName":"","lastName":"Chauvie","suffix":""},{"id":623395656,"identity":"ff2f1719-8dae-4f19-bcea-b5892c19e451","order_by":4,"name":"Dominik Amereller","email":"","orcid":"","institution":"Department of Biomedical Imaging and Image-Guided Therapy, Division of Nuclear Medicine, Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Dominik","middleName":"","lastName":"Amereller","suffix":""},{"id":623395657,"identity":"0d02dff4-ba79-4abc-87ae-18117473a067","order_by":5,"name":"Elisabetta Abenavoli","email":"","orcid":"","institution":"Division of Nuclear Medicine, Azienda Ospedaliero Universitaria Careggi, Florence, Italy","correspondingAuthor":false,"prefix":"","firstName":"Elisabetta","middleName":"","lastName":"Abenavoli","suffix":""},{"id":623395658,"identity":"51b80544-fc8c-4b68-867d-1ed5e3ea25fd","order_by":6,"name":"Aleksa Lazarevic","email":"","orcid":"","institution":"Department of Biomedical Imaging and Image-Guided Therapy, Division of Nuclear Medicine, Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Aleksa","middleName":"","lastName":"Lazarevic","suffix":""},{"id":623395659,"identity":"ef3133b0-1e1d-40c8-9244-8853346e5e0e","order_by":7,"name":"Natalia Wegrzyn","email":"","orcid":"","institution":"Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Natalia","middleName":"","lastName":"Wegrzyn","suffix":""},{"id":623395660,"identity":"1303fc6d-1fd2-4c3b-8e0f-8362298464fb","order_by":8,"name":"Philip Richter","email":"","orcid":"","institution":"Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Philip","middleName":"","lastName":"Richter","suffix":""},{"id":623395661,"identity":"d0a03276-f10a-4546-888b-eb4410a8a185","order_by":9,"name":"Iustin Tibu","email":"","orcid":"","institution":"Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Iustin","middleName":"","lastName":"Tibu","suffix":""},{"id":623395662,"identity":"c7ab8ffd-de55-4397-945a-8540d45e8a03","order_by":10,"name":"Ken Miller","email":"","orcid":"","institution":"Department of Biomedical Imaging and Image-Guided Therapy, Division of Nuclear Medicine, Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Ken","middleName":"","lastName":"Miller","suffix":""},{"id":623395663,"identity":"a21b28e6-d94e-4194-872d-5d3a3726643b","order_by":11,"name":"David Kersting","email":"","orcid":"","institution":"Department of Nuclear Medicine, West German Cancer Center (WTZ), University Hospital Essen, University of Duisburg-Essen, Essen, Germany German Cancer Consortium (DKTK), Partner Site University Hospital Essen, Essen, Germany; German Cancer Consortium (DKTK), Partner Site University Hospital Essen, Essen, Germany","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Kersting","suffix":""},{"id":623395664,"identity":"2bc7d0ad-c919-4305-bed3-a6f6bf5191f6","order_by":12,"name":"Moon-Sung Kim","email":"","orcid":"","institution":"Institute for AI in Medicine (IKIM), University Hospital Essen, Essen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Moon-Sung","middleName":"","lastName":"Kim","suffix":""},{"id":623395665,"identity":"a9d9855f-46b4-48ab-84c2-5cbdb60134d2","order_by":13,"name":"Jakob Dexl","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany; Munich Center for Machine Learning (MCML), Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jakob","middleName":"","lastName":"Dexl","suffix":""},{"id":623395666,"identity":"882de261-cd72-43f9-9864-1287e6406d60","order_by":14,"name":"Michael Ingrisch","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany; Munich Center for Machine Learning (MCML), Munich, Germany; relAI – Konrad Zuse School of Excellence in Reliable AI, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Ingrisch","suffix":""},{"id":623395667,"identity":"372ff266-53fe-432d-8b1b-681fefbcbb43","order_by":15,"name":"Clemens C. Cyran","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Clemens","middleName":"C.","lastName":"Cyran","suffix":""},{"id":623395668,"identity":"fe0744fc-1569-45fb-8aca-6ab08aee6c84","order_by":16,"name":"Maurice Heimer","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Maurice","middleName":"","lastName":"Heimer","suffix":""},{"id":623395669,"identity":"2131b236-f7a3-49ce-a361-0884b5520c7d","order_by":17,"name":"Mej Kazazi","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Mej","middleName":"","lastName":"Kazazi","suffix":""},{"id":623395670,"identity":"e706d34f-e41e-41fb-ae6c-01c7e7be17db","order_by":18,"name":"Jason Callahan","email":"","orcid":"","institution":"Department of Cancer Imaging, Peter MacCallum Cancer Centre, Melbourne, VIC, Australia","correspondingAuthor":false,"prefix":"","firstName":"Jason","middleName":"","lastName":"Callahan","suffix":""},{"id":623395671,"identity":"4a635b16-745c-4eab-8106-5c8ce33b469f","order_by":19,"name":"Rodney Hicks","email":"","orcid":"","institution":"Department of Cancer Imaging, Peter MacCallum Cancer Centre, Melbourne, VIC, Australia St Vincent's Hospital, Department of Medicine, The University of Melbourne, Fitzroy, Victoria 3065, Australia; St Vincent's Hospital, Department of Medicine, The University of Melbourne, Fitzroy, Victoria 3065, Australia","correspondingAuthor":false,"prefix":"","firstName":"Rodney","middleName":"","lastName":"Hicks","suffix":""},{"id":623395672,"identity":"17cd645d-2fba-46fd-b2ef-a47fc7cd0f71","order_by":20,"name":"Eduardo Ochoa","email":"","orcid":"","institution":"Department of Nuclear Medicine and Clinical Molecular Imaging, University Hospital Tubingen, Tubingen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Eduardo","middleName":"","lastName":"Ochoa","suffix":""},{"id":623395673,"identity":"fb6a6d3d-0d62-4129-955c-41c24043da07","order_by":21,"name":"Jonas Benecke","email":"","orcid":"","institution":"Department of Nuclear Medicine and Clinical Molecular Imaging, University Hospital Tubingen, Tubingen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Jonas","middleName":"","lastName":"Benecke","suffix":""},{"id":623395674,"identity":"70bbe89e-59ec-4398-aa7a-71f82d8141ec","order_by":22,"name":"Fabian Schmidt","email":"","orcid":"","institution":"Department of Nuclear Medicine and Clinical Molecular Imaging, University Hospital Tubingen, Tubingen, Germany; Werner Siemens Imaging Center, Department of Preclinical Imaging and Radiopharmacy, Eberhard Karls University Tubingen, Tubingen, Germany","correspondingAuthor":false,"prefix":"","firstName":"Fabian","middleName":"","lastName":"Schmidt","suffix":""},{"id":623395675,"identity":"9cd8ac46-4462-44c4-820e-a3f066de7b20","order_by":23,"name":"Roberto Sciagra","email":"","orcid":"","institution":"Division of Nuclear Medicine, Azienda Ospedaliero Universitaria Careggi, Florence, Italy","correspondingAuthor":false,"prefix":"","firstName":"Roberto","middleName":"","lastName":"Sciagra","suffix":""},{"id":623395676,"identity":"67699a82-c87b-41d3-a336-20badd4adb3c","order_by":24,"name":"Julie Auriac","email":"","orcid":"","institution":"IRIS, Institut Curie, Inserm, CNRS, Université Versailles Saint-Quentin, Universite PSL, Orsay, France","correspondingAuthor":false,"prefix":"","firstName":"Julie","middleName":"","lastName":"Auriac","suffix":""},{"id":623395677,"identity":"87b6423c-0556-4112-9a03-88854c8f9bb0","order_by":25,"name":"Romain Seban-David","email":"","orcid":"","institution":"IRIS, Institut Curie, Inserm, CNRS, Université Versailles Saint-Quentin, Universite PSL, Orsay, France","correspondingAuthor":false,"prefix":"","firstName":"Romain","middleName":"","lastName":"Seban-David","suffix":""},{"id":623395678,"identity":"b32d8920-b4fa-44f2-baf4-a46d72916dd3","order_by":26,"name":"Fanny Orlhac","email":"","orcid":"","institution":"IRIS, Institut Curie, Inserm, CNRS, Université Versailles Saint-Quentin, Universite PSL, Orsay, France","correspondingAuthor":false,"prefix":"","firstName":"Fanny","middleName":"","lastName":"Orlhac","suffix":""},{"id":623395679,"identity":"4fbb610e-aba1-4b58-9e48-609b72889a37","order_by":27,"name":"Irene Buvat","email":"","orcid":"","institution":"IRIS, Institut Curie, Inserm, CNRS, Université Versailles Saint-Quentin, Universite PSL, Orsay, France","correspondingAuthor":false,"prefix":"","firstName":"Irene","middleName":"","lastName":"Buvat","suffix":""},{"id":623395680,"identity":"dcb93a98-0caa-4c80-b00b-d27c0fadf437","order_by":28,"name":"Sophie C. Siegmund","email":"","orcid":"","institution":"Department of Nuclear Medicine, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Sophie","middleName":"C.","lastName":"Siegmund","suffix":""},{"id":623395681,"identity":"53a6b453-af53-4ce2-923b-4461e092bd39","order_by":29,"name":"Liam Widjaja","email":"","orcid":"","institution":"Department of Nuclear Medicine, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Liam","middleName":"","lastName":"Widjaja","suffix":""},{"id":623395682,"identity":"e32f38cb-08df-4e2e-9e9c-4ca996384f0a","order_by":30,"name":"Rudolf A. Werner","email":"","orcid":"","institution":"Department of Nuclear Medicine, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Rudolf","middleName":"A.","lastName":"Werner","suffix":""},{"id":623395683,"identity":"a1963dbf-3353-4e18-8616-299561db3ec0","order_by":31,"name":"Giulia Santo","email":"","orcid":"","institution":"Department of Nuclear Medicine, Medical University of Innsbruck, Innsbruck, Austria","correspondingAuthor":false,"prefix":"","firstName":"Giulia","middleName":"","lastName":"Santo","suffix":""},{"id":623395684,"identity":"e6c85205-b9de-481b-af0c-b0c0bc451c23","order_by":32,"name":"Ariane Kronthaler","email":"","orcid":"","institution":"Department of Nuclear Medicine, Medical University of Innsbruck, Innsbruck, Austria","correspondingAuthor":false,"prefix":"","firstName":"Ariane","middleName":"","lastName":"Kronthaler","suffix":""},{"id":623395685,"identity":"c4ac4af1-d8b9-49dc-8808-586ad2d908bd","order_by":33,"name":"Irene J. Virgolini","email":"","orcid":"","institution":"Department of Nuclear Medicine, Medical University of Innsbruck, Innsbruck, Austria","correspondingAuthor":false,"prefix":"","firstName":"Irene","middleName":"J.","lastName":"Virgolini","suffix":""},{"id":623395686,"identity":"b0c69798-f435-4693-994d-a5273f848f03","order_by":34,"name":"Yasser G. Abdelhafez","email":"","orcid":"","institution":"Department of Radiology, University of California, Davis, Davis, California, USA; Nuclear Medicine Unit, South Egypt Cancer Institute, Assiut University, Egypt","correspondingAuthor":false,"prefix":"","firstName":"Yasser","middleName":"G.","lastName":"Abdelhafez","suffix":""},{"id":623395687,"identity":"79463239-a3fc-4603-8211-cf32b29b2c90","order_by":35,"name":"Hande Nalbant","email":"","orcid":"","institution":"Department of Radiology, University of California, Davis, Davis, California, USA","correspondingAuthor":false,"prefix":"","firstName":"Hande","middleName":"","lastName":"Nalbant","suffix":""},{"id":623395688,"identity":"146a6940-82be-4cf8-84d9-466ca91f909a","order_by":36,"name":"Miju Cheon","email":"","orcid":"","institution":"Department of Radiology, University of California, Davis, Davis, California, USA; Department of Nuclear Medicine, Veterans Health Service Medical Center, Seoul, Korea","correspondingAuthor":false,"prefix":"","firstName":"Miju","middleName":"","lastName":"Cheon","suffix":""},{"id":623395689,"identity":"f82cd55a-2cab-4ef1-a541-7e5e174e4598","order_by":37,"name":"Ramsey D. Badawi","email":"","orcid":"","institution":"Department of Radiology, University of California, Davis, Davis, California, USA","correspondingAuthor":false,"prefix":"","firstName":"Ramsey","middleName":"D.","lastName":"Badawi","suffix":""},{"id":623395690,"identity":"637cec82-8316-4753-8a3d-5de929f865b1","order_by":38,"name":"Anting Li","email":"","orcid":"","institution":"Turku PET Centre, University of Turku and Turku University Hospital, Turku, Finland","correspondingAuthor":false,"prefix":"","firstName":"Anting","middleName":"","lastName":"Li","suffix":""},{"id":623395691,"identity":"60d185f9-81bb-4baf-bf6e-97bcaa0b4e2f","order_by":39,"name":"Ahti Vehmanen","email":"","orcid":"","institution":"Turku PET Centre, University of Turku and Turku University Hospital, Turku, Finland; Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital, Turku, Finland","correspondingAuthor":false,"prefix":"","firstName":"Ahti","middleName":"","lastName":"Vehmanen","suffix":""},{"id":623395692,"identity":"35ddb7d3-ed04-4661-9d48-688c235813c1","order_by":40,"name":"Jukka Kemppainen","email":"","orcid":"","institution":"Turku PET Centre, University of Turku and Turku University Hospital, Turku, Finland; Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital, Turku, Finland","correspondingAuthor":false,"prefix":"","firstName":"Jukka","middleName":"","lastName":"Kemppainen","suffix":""},{"id":623395693,"identity":"7452b7ef-1087-403d-b2dd-d93dfdeb8ebe","order_by":41,"name":"Riku Klén","email":"","orcid":"","institution":"Turku PET Centre, University of Turku and Turku University Hospital, Turku, Finland","correspondingAuthor":false,"prefix":"","firstName":"Riku","middleName":"","lastName":"Klén","suffix":""},{"id":623395694,"identity":"fcf10339-3dd7-48c8-a858-7a2220331f95","order_by":42,"name":"Thomas Beyer","email":"","orcid":"","institution":"QIMP Team, Medical University of Vienna, Vienna, Austria","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Beyer","suffix":""},{"id":623395695,"identity":"0aeb5976-8cf5-45ee-acb7-1f321cd8f8b7","order_by":43,"name":"Lalith K. S. Sundar","email":"","orcid":"","institution":"Department of Radiology, LMU University Hospital, LMU Munich, Munich, Germany","correspondingAuthor":false,"prefix":"","firstName":"Lalith","middleName":"K. S.","lastName":"Sundar","suffix":""}],"badges":[],"createdAt":"2026-04-13 18:34:34","currentVersionCode":2,"declarations":{"humanSubjects":true,"vertebrateSubjects":false,"conflictsOfInterestStatement":true,"humanSubjectEthicalGuidelines":true,"humanSubjectConsent":true,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9407530/v2","doiUrl":"https://doi.org/10.21203/rs.3.rs-9407530/v2","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107869914,"identity":"4d36a566-edbd-41c7-940f-615eba9cbda3","added_by":"auto","created_at":"2026-04-27 07:38:26","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":901253,"visible":true,"origin":"","legend":"","description":"","filename":"LIONmanuscriptfull.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9407530/v2_covered_7496131b-4a24-4350-b17e-8cb83ca76200.pdf"},{"id":107750107,"identity":"4f96708a-6ff6-4d57-87a9-14a605c7a955","added_by":"auto","created_at":"2026-04-24 17:03:03","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":195815,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-9407530/v2/f4ec414349a4a4251de066bf.docx"}],"financialInterests":"The authors declare potential competing interests as follows: Lalith Kumar Shiyam Sundar and Thomas Beyer are cofounders of Zenta GmbH/Austria. D. Kersting reports a research grant from Pfizer, personal fees from Alnylam, GE Healthcare, Novartis, and Pfizer, and travel grants from Life Molecular Imaging and Sanofi, all outside of the submitted work. Rudolf A. Werner has received speaker honoraria from Novartis.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eFully automated segmentation of [\u003c/strong\u003e\u003csup\u003e\u003cstrong\u003e18\u003c/strong\u003e\u003c/sup\u003e\u003cstrong\u003eF]FDG- and PSMA-PET/CT images via data-centric Deep-Learning.\u003c/strong\u003e\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"FWF Austrian Science Fund","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":"","lastPublishedDoi":"10.21203/rs.3.rs-9407530/v2","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9407530/v2","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003ePurpose \u003c/strong\u003eTo develop and validate LION (Lesion Identification in Oncological Nuclear imaging), an open-source PET-only tumor segmentation pipeline for [\u003csup\u003e18\u003c/sup\u003eF]FDG and PSMA-targeted PET/CT, and to investigate how training data characteristics influence segmentation performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterials and Methods \u003c/strong\u003eIn this retrospective multicenter study, 5,209 [\u003csup\u003e18\u003c/sup\u003eF]FDG PET/CT scans spanning 19 disease types and 2,046 PSMA-targeted PET/CT scans were used to train PET-only segmentation models. Tumor segmentation incorporated organs with physiological uptake as auxiliary classes to enable PET-only inference. Tumor Occurrence Maps (TOMs) quantified tumor spatial diversity across the training data. For [\u003csup\u003e18\u003c/sup\u003eF]FDG, disease-specific and mixed-disease models trained on progressively larger subsets were compared to test whether increasing spatial diversity improves generalization. Scanner-related domain shift was analyzed using DINOv2 embeddings. Models were evaluated on multicenter holdout cohorts (616 [\u003csup\u003e18\u003c/sup\u003eF]FDG across 4 diseases; 443 PSMA-targeted prostate cancer scans) and compared with three open-source tools.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults \u003c/strong\u003eOrgan context improved median Dice from 0.62 to 0.71 for [\u003csup\u003e18\u003c/sup\u003eF]FDG and from 0.75 to 0.83 for PSMA. Spatial diversity measured by TOMs was strongly associated with Dice (Spearman ρ = 0.80, P = 0.003). A mixed-disease model trained on 500 patients matched the performance of a lymphoma specialist model trained on 3,031 cases. DINOv2 embeddings revealed scanner-induced domain shift, between same-disease cohorts. LION achieved median Dice scores of 0.71 ([\u003csup\u003e18\u003c/sup\u003eF]FDG) and 0.85 (PSMA) and outperformed other open-source approaches on common holdout patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion \u003c/strong\u003eLION enables PET-only automated segmentation for [\u003csup\u003e18\u003c/sup\u003eF]FDG and PSMA-targeted PET. Training data composition, particularly spatial diversity quantified by TOMs, was strongly associated with segmentation performance.\u003c/p\u003e","manuscriptTitle":"Fully automated segmentation of [18F]FDG- and PSMA-PET/CT images via data-centric Deep-Learning.","msid":"","msnumber":"","nonDraftVersions":[{"code":2,"date":"2026-04-24 17:02:59","doi":"10.21203/rs.3.rs-9407530/v2","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}},{"code":1,"date":"2026-04-15 07:04:11","doi":"10.21203/rs.3.rs-9407530/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":"05184247-cb5b-49a7-ae7a-e8cea8365d79","owner":[],"postedDate":"April 24th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":66969929,"name":"Nuclear Medicine \u0026 Medical Imaging"},{"id":66969930,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-04-15T07:04:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-24 17:02:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v2","identity":"rs-9407530","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9407530","identity":"rs-9407530","version":["v2"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.