A Two-Stage Deep Learning Pipeline for Gamma Radiation-Induced Dicentric Chromosome classification Using YOLOv8 & ResNet18

A Two-Stage Deep Learning Pipeline for Gamma Radiation-Induced Dicentric Chromosome classification Using YOLOv8 & ResNet18 | 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 A Two-Stage Deep Learning Pipeline for Gamma Radiation-Induced Dicentric Chromosome classification Using YOLOv8 & ResNet18 Rishabh Mohan Sinha, Namita Indracanti, Prem Indraganti This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9297818/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The ability to detect biological damage caused by radiation exposure rapidly is a core part of Chemical, Biological, Radiological, and Nuclear (CBRN) defence readiness. Internationally, dicentric chromosomes (DCs) resulting from the misrepair of double-stranded DNA breaks are the most accepted, dependable cytogenetic biomarker of ionizing radiation exposure. However, existing automated approaches are limited by a lack of interpretability, dependence on end-to-end architectures, and insufficient robustness for real-world deployment. Still, the conventional Dicentric Chromosome Assay (DCA) is labor intensive, qualitative, and impracticable for high-volume triage applications in battlefield or nuclear accident scenarios. In this paper we present a two-stage deep learning pipeline that automates cytogenetic biodosimetry and is relevant for defence applications. In the first stage, object detection using a YOLOv8 model located individual chromosomes robustly from metaphase spreads. In the second stage, a lightweight ResNet18 classifier determined whether cropped chromosomes were normal or dicentric. Overall, the pipeline was trained on a curated dataset of 325 metaphase plates (~13,000 chromosomes). The results of the automated pipeline were validation outcomes listed as follows 0.92 precision, recall of 0.94, [email protected] of 0.95, and overall accuracy classification of 94.2%. Our proposed system tremendously decreases the analysis time from hours to seconds, which enables rapid triage and diagnosis of radiation exposure for defence personnel and, by extension, civilian populations. The entire analysis pipeline is modular and can be deployed flexibly on lightweight hardware, which is ideal for use in, but not limited to, field-laboratories, mobile biodosimetry units, and hospitals. This work can be translated AI-driven cytogenetics into defence biosurveillance and can provide a deployable solution for live & accurate monitoring in the incident of nuclear accidents, radiological terrorism or in the event of radiation exposure on the battlefield. Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted 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-9297818","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":620973267,"identity":"9eddd58e-bbbe-467b-ae99-63eb1f97625d","order_by":0,"name":"Rishabh Mohan Sinha","email":"data:image/png;base64,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","orcid":"","institution":"Delhi Technological University","correspondingAuthor":true,"prefix":"","firstName":"Rishabh","middleName":"Mohan","lastName":"Sinha","suffix":""},{"id":620973268,"identity":"920a41e8-439a-4b68-bc09-1d7f789599ef","order_by":1,"name":"Namita Indracanti","email":"","orcid":"","institution":"Institute of Nuclear Medicine \u0026 Allied Sciences","correspondingAuthor":false,"prefix":"","firstName":"Namita","middleName":"","lastName":"Indracanti","suffix":""},{"id":620973269,"identity":"6e0e8964-932a-43fa-b161-d432b016d226","order_by":2,"name":"Prem Indraganti","email":"","orcid":"","institution":"Institute of Nuclear Medicine \u0026 Allied Sciences","correspondingAuthor":false,"prefix":"","firstName":"Prem","middleName":"","lastName":"Indraganti","suffix":""}],"badges":[],"createdAt":"2026-04-02 04:08:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9297818/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9297818/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107484871,"identity":"61e2a9d1-5c21-4da8-ae5e-b99202de379f","added_by":"auto","created_at":"2026-04-22 02:33:13","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":399178,"visible":true,"origin":"","legend":"","description":"","filename":"RISHABHDRDOPUBLICATIONCopy11.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9297818/v1_covered_9cc1102b-8938-4443-a652-70ee547080ae.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Two-Stage Deep Learning Pipeline for Gamma Radiation-Induced Dicentric Chromosome classification Using YOLOv8 \u0026amp; ResNet18","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"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-9297818/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9297818/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The ability to detect biological damage caused by radiation exposure rapidly is a core part of Chemical, Biological, Radiological, and Nuclear (CBRN) defence readiness. Internationally, dicentric chromosomes (DCs) resulting from the misrepair of double-stranded DNA breaks are the most accepted, dependable cytogenetic biomarker of ionizing radiation exposure. However, existing automated approaches are limited by a lack of interpretability, dependence on end-to-end architectures, and insufficient robustness for real-world deployment. Still, the conventional Dicentric Chromosome Assay (DCA) is labor intensive, qualitative, and impracticable for high-volume triage applications in battlefield or nuclear accident scenarios. In this paper we present a two-stage deep learning pipeline that automates cytogenetic biodosimetry and is relevant for defence applications. In the first stage, object detection using a YOLOv8 model located individual chromosomes robustly from metaphase spreads. In the second stage, a lightweight ResNet18 classifier determined whether cropped chromosomes were normal or dicentric. Overall, the pipeline was trained on a curated dataset of 325 metaphase plates (~13,000 chromosomes). The results of the automated pipeline were validation outcomes listed as follows 0.92 precision, recall of 0.94, [email protected] of 0.95, and overall accuracy classification of 94.2%. Our proposed system tremendously decreases the analysis time from hours to seconds, which enables rapid triage and diagnosis of radiation exposure for defence personnel and, by extension, civilian populations. The entire analysis pipeline is modular and can be deployed flexibly on lightweight hardware, which is ideal for use in, but not limited to, field-laboratories, mobile biodosimetry units, and hospitals.\nThis work can be translated AI-driven cytogenetics into defence biosurveillance and can provide a deployable solution for live \u0026 accurate monitoring in the incident of nuclear accidents, radiological terrorism or in the event of radiation exposure on the battlefield.","manuscriptTitle":"A Two-Stage Deep Learning Pipeline for Gamma Radiation-Induced Dicentric Chromosome classification Using YOLOv8 \u0026amp; ResNet18","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-20 15:18:04","doi":"10.21203/rs.3.rs-9297818/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":"6f60bc8a-4bee-4aef-9902-bdcf97eca2c7","owner":[],"postedDate":"April 20th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-20T15:18:04+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-20 15:18:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9297818","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9297818","identity":"rs-9297818","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}