ThoraxSense: Enhanced Thoracic Multi-DiseaseDetection on Chest X-Rays Using DenseNet121 andClass-Imbalance Optimization

ThoraxSense: Enhanced Thoracic Multi-DiseaseDetection on Chest X-Rays Using DenseNet121 andClass-Imbalance Optimization | 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 ThoraxSense: Enhanced Thoracic Multi-DiseaseDetection on Chest X-Rays Using DenseNet121 andClass-Imbalance Optimization Ansh Srivastava, Rishi Raj, Rishit Sharma, Rayan Haque, Rashmi K.B, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8277163/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 Thoracic diseases are commonly detected through Chest X-rays, however class imbalance, label noise, computationalconstraints in large clinical datasets obstruct automated accurate interpretation. To address these challenges, we introduce ThoraxSense, a resource-efficient framework for multi-disease classification. In the first implementation, we utilize a PyTorch-based DenseNet121 backbone integrated with targeted GPU optimizations consisting of CUDA-aware memory management, adaptive learning-rate scheduling, and gradient clipping which in turn resulted in improved training stability on restrictedhardware. In order to achieve robust convergence and stable results, we apply dynamic class-imbalance compensationusing weighted loss functions. Further, a TensorFlow/Keras pipeline using a fine-tuned VGG16 architecture was developed inorder to evaluate cross-framework consistency. This allowed a comparative analysis across deep learning ecosystems. TheKeras-based VGG16 model achieved a mean AUROC of 0.8003 and micro AUROC of 0.8406. Focusing on reliable learning,hardware-efficient optimization, and reproducible cross-framework performance, ThoraxSense emerges as a reproducible anda practically deployable solution for thoracic disease detection, supporting real-world clinical needs. Biological sciences/Computational biology and bioinformatics Physical sciences/Engineering Physical sciences/Mathematics and computing Class Imbalance DenseNet121 CUDA Gradient Clipping VGG16 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-8277163","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":583206537,"identity":"0c77d7fd-8881-464b-abdb-c005bf93b611","order_by":0,"name":"Ansh Srivastava","email":"","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":false,"prefix":"","firstName":"Ansh","middleName":"","lastName":"Srivastava","suffix":""},{"id":583206538,"identity":"63c80f01-be95-47c0-827f-6ddb8253bf08","order_by":1,"name":"Rishi Raj","email":"","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":false,"prefix":"","firstName":"Rishi","middleName":"","lastName":"Raj","suffix":""},{"id":583206539,"identity":"73fc600d-b233-417a-ad3b-690e02d6d785","order_by":2,"name":"Rishit Sharma","email":"","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":false,"prefix":"","firstName":"Rishit","middleName":"","lastName":"Sharma","suffix":""},{"id":583206540,"identity":"f829e8d2-af76-4a7d-a90f-59a5f6a8e30a","order_by":3,"name":"Rayan Haque","email":"","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":false,"prefix":"","firstName":"Rayan","middleName":"","lastName":"Haque","suffix":""},{"id":583206541,"identity":"b1465ba4-1816-4a01-ade8-8ceff4957de0","order_by":4,"name":"Rashmi K.B","email":"","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":false,"prefix":"","firstName":"Rashmi","middleName":"","lastName":"K.B","suffix":""},{"id":583206542,"identity":"de8d4955-0ea6-4d5f-a67f-485f6e5202bf","order_by":5,"name":"Shobana T.S","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCUlEQVRIiWNgGAWjYDACCQY2KIuxgYGh4gBcgoeQFgmIljOkaQFZ1HYAp0I4kJ/dfOzBzx12dfINzM0fPs67Ey3ffvgAw48aBhlzHFoM7hxLN+w9kyxhcICxTXLmtme5G86kJTD2HGPgsWzAoUUix0yCt41ZwgDoKmbebYdzN0jwGDDwNjDwGOBwpPyMHDPJv231EvINjM2f/845nDt/Bv8Hxr94tDDcyDGT5m07LMFwgLFBmrHhcG7DDR4GZny2GNxISzeWbTsuueEw0C89x8B+MTgsc0wCj8OSjz1821bNL9/e/vjDj5o7ufPbDz98+KbGxh6nw+CAGYl9ABZRo2AUjIJRMArIAwBqzFx8unWYkwAAAABJRU5ErkJggg==","orcid":"","institution":"Visvesvaraya Technological University","correspondingAuthor":true,"prefix":"","firstName":"Shobana","middleName":"","lastName":"T.S","suffix":""}],"badges":[],"createdAt":"2025-12-04 09:08:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8277163/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8277163/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106413243,"identity":"c85d35de-8c9e-4452-9615-d78d590217cd","added_by":"auto","created_at":"2026-04-08 10:03:37","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1132550,"visible":true,"origin":"","legend":"","description":"","filename":"ThoraxSenseEnhancedThoracicMultiDiseaseDetectiononChestXRaysUsingDenseNet121andClassImbalanceOptimization.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8277163/v1_covered_bb2eae46-db30-4674-9192-5ef00fbef5f4.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"ThoraxSense: Enhanced Thoracic Multi-DiseaseDetection on Chest X-Rays Using DenseNet121 andClass-Imbalance Optimization","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"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":"Class Imbalance, DenseNet121, CUDA, Gradient Clipping, VGG16","lastPublishedDoi":"10.21203/rs.3.rs-8277163/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8277163/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Thoracic diseases are commonly detected through Chest X-rays, however class imbalance, label noise, computationalconstraints in large clinical datasets obstruct automated accurate interpretation. To address these challenges, we introduce\nThoraxSense, a resource-efficient framework for multi-disease classification. In the first implementation, we utilize a PyTorch-based DenseNet121 backbone integrated with targeted GPU optimizations consisting of CUDA-aware memory management,\nadaptive learning-rate scheduling, and gradient clipping which in turn resulted in improved training stability on restrictedhardware. In order to achieve robust convergence and stable results, we apply dynamic class-imbalance compensationusing weighted loss functions. Further, a TensorFlow/Keras pipeline using a fine-tuned VGG16 architecture was developed inorder to evaluate cross-framework consistency. This allowed a comparative analysis across deep learning ecosystems. TheKeras-based VGG16 model achieved a mean AUROC of 0.8003 and micro AUROC of 0.8406. Focusing on reliable learning,hardware-efficient optimization, and reproducible cross-framework performance, ThoraxSense emerges as a reproducible anda practically deployable solution for thoracic disease detection, supporting real-world clinical needs.","manuscriptTitle":"ThoraxSense: Enhanced Thoracic Multi-DiseaseDetection on Chest X-Rays Using DenseNet121 andClass-Imbalance Optimization","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-01 14:06:57","doi":"10.21203/rs.3.rs-8277163/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":"97293b2d-8003-49f0-b2f0-bb6c41f526a4","owner":[],"postedDate":"February 1st, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":62052839,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":62052840,"name":"Physical sciences/Engineering"},{"id":62052841,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-04-08T09:50:35+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-01 14:06:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8277163","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8277163","identity":"rs-8277163","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}