WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical Imaging

WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical Imaging | 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 WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical Imaging Rao Farhat Masood, Imtiaz Ahmad Taj This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9019468/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 Class imbalance in medical imaging datasets remains a key challenge for reliable object detection, particularly when rare yet clinically significant pathologies coexist with prevalent findings. In spinal MRI, common conditions such as Normal Intervertebral Disc (IVD) may constitute over 45% of annotated objects, whereas findings like Spondylolisthesis account for fewer than 2% of instances. Conventional loss functions including Focal Loss, Class-Balanced Loss, and Label-Distribution-Aware Margin Loss, each address isolated facets of this imbalance but do not provide a unified, adaptive solution. Inspired by ensemble loss strategies recently advanced in Deep Metric Learning (DML), we propose WELDE ( W eighted E nsemble L oss with D iversity E nhancement), a framework that combines four complementary loss functions via per-head adapter projections, EMA-based normalization, and learnable adaptive weighting with a relaxed sum-to-one penalty. Each loss component receives a dedicated classification head with an independent adapter projection from a shared frozen backbone, enabling feature specialization without backbone fine-tuning. We provide theoretical analysis of WELDE's properties, including gradient magnitude balancing across loss components and weight non-degeneracy. Applied to a lumbar mid-sagittal spinal MRI dataset with six classes and a 33.9:1 imbalance ratio, WELDE achieves the highest classification performance among all evaluated methods, outperforming all single-loss baselines (mAP 0.702 vs.\0.689 for the best baseline CE, mAP \((_{\text{tail}})\) 0.509 vs.\0.472, \((+)\) 8.1% relative improvement on tail classes) and an architecture-matched CE ensemble control (mAP \((_{\text{tail}})\) 0.509 vs.\0.496), confirming that the improvement derives from diverse loss composition rather than increased model capacity. External cross-domain validation on the DermaMNIST skin lesion benchmark (7 classes, \((\rho{=}58.3)\) ) confirms that \welde{} generalizes robustly, achieving the highest mAP ( \((0.709)\) ) and mAP \((_{\text{tail}})\) ( \((0.651)\) ) among all methods, outperforming both single-head baselines ( \((+11.5%)\) mAP over CE) and the architecture-matched CE ensemble control. class imbalance object detection ensemble loss deep metric learning spinal MRI per-head adapters medical imaging cross-domain validation 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-9019468","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":609979321,"identity":"c1d17615-aa6f-401e-a075-56e6e4dd1a27","order_by":0,"name":"Rao Farhat Masood","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBElEQVRIiWNgGAWjYBAC9gYGNjhHgsGGQQ7EOPAAjxaeAyAtCTAtaQzGYC0JpGhJbACx8GphP/vswc8f2+Tl25sP3mBI2JY+P+zwQ6AtdnK6DTi08KSbG/Yk3DbccOZYsgVDwu3cjbfTDIBako3NDmDXYs+QxibBk3CbcYNEjpkE4w+gltkJIC0HErfh0MLD/4xN8k/Cbfv5M4BagLakG85O/4Bfi0QamzTQlsSGGxAtCfLSOQRskXjGJi2TdjsZ7JcEkKekcwoOJBjg9gsPfxqb5Bub27bzQSH2IeG2vPzs9M0fPlTYyeHSggoSgNgArNKAGOUwIN9AiupRMApGwSgYCQAAsm5iRGwniT0AAAAASUVORK5CYII=","orcid":"","institution":"Capital University of Science and Technology (CUST)","correspondingAuthor":true,"prefix":"","firstName":"Rao","middleName":"Farhat","lastName":"Masood","suffix":""},{"id":609979322,"identity":"b09903c9-d9dd-419e-9c47-a1465262b904","order_by":1,"name":"Imtiaz Ahmad Taj","email":"","orcid":"","institution":"Capital University of Science and Technology (CUST)","correspondingAuthor":false,"prefix":"","firstName":"Imtiaz","middleName":"Ahmad","lastName":"Taj","suffix":""}],"badges":[],"createdAt":"2026-03-03 11:09:41","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9019468/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9019468/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105564851,"identity":"2105c5ce-74c0-405c-88e4-6f71ab2d54d9","added_by":"auto","created_at":"2026-03-27 12:51:05","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1681425,"visible":true,"origin":"","legend":"","description":"","filename":"WELDE.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9019468/v1_covered_6778dd99-5366-4dad-944b-9fd4b1d9f9ce.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical Imaging","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":"class imbalance, object detection, ensemble loss, deep metric learning, spinal MRI, per-head adapters, medical imaging, cross-domain validation","lastPublishedDoi":"10.21203/rs.3.rs-9019468/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9019468/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eClass imbalance in medical imaging datasets remains a key challenge for reliable object detection, particularly when rare yet clinically significant pathologies coexist with prevalent findings. In spinal MRI, common conditions such as Normal Intervertebral Disc (IVD) may constitute over 45% of annotated objects, whereas findings like Spondylolisthesis account for fewer than 2% of instances. Conventional loss functions including Focal Loss, Class-Balanced Loss, and Label-Distribution-Aware Margin Loss, each address isolated facets of this imbalance but do not provide a unified, adaptive solution. Inspired by ensemble loss strategies recently advanced in Deep Metric Learning (DML), we propose \u003cb\u003eWELDE\u003c/b\u003e (\u003cb\u003eW\u003c/b\u003eeighted \u003cb\u003eE\u003c/b\u003ensemble \u003cb\u003eL\u003c/b\u003eoss with \u003cb\u003eD\u003c/b\u003eiversity \u003cb\u003eE\u003c/b\u003enhancement), a framework that combines four complementary loss functions via per-head adapter projections, EMA-based normalization, and learnable adaptive weighting with a relaxed sum-to-one penalty. Each loss component receives a dedicated classification head with an independent adapter projection from a shared frozen backbone, enabling feature specialization without backbone fine-tuning. We provide theoretical analysis of WELDE's properties, including gradient magnitude balancing across loss components and weight non-degeneracy. Applied to a lumbar mid-sagittal spinal MRI dataset with six classes and a 33.9:1 imbalance ratio, WELDE achieves the highest classification performance among all evaluated methods, outperforming all single-loss baselines (mAP 0.702 vs.\\0.689 for the best baseline CE, mAP\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((_{\\text{tail}})\\)\u003c/span\u003e\u003c/span\u003e 0.509 vs.\\0.472, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((+)\\)\u003c/span\u003e\u003c/span\u003e8.1% relative improvement on tail classes) and an architecture-matched CE ensemble control (mAP\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((_{\\text{tail}})\\)\u003c/span\u003e\u003c/span\u003e 0.509 vs.\\0.496), confirming that the improvement derives from diverse loss composition rather than increased model capacity. External cross-domain validation on the DermaMNIST skin lesion benchmark (7 classes, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((\\rho{=}58.3)\\)\u003c/span\u003e\u003c/span\u003e) confirms that \\welde{} generalizes robustly, achieving the highest mAP (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((0.709)\\)\u003c/span\u003e\u003c/span\u003e) and mAP\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((_{\\text{tail}})\\)\u003c/span\u003e\u003c/span\u003e (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((0.651)\\)\u003c/span\u003e\u003c/span\u003e) among all methods, outperforming both single-head baselines (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((+11.5%)\\)\u003c/span\u003e\u003c/span\u003e mAP over CE) and the architecture-matched CE ensemble control.\u003c/p\u003e","manuscriptTitle":"WELDE: A Weighted Ensemble Loss with Diversity Enhancement for Imbalanced Object Detection in Medical Imaging","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-24 13:51:47","doi":"10.21203/rs.3.rs-9019468/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":"a014dadd-24cf-4ff6-8360-82cd96997258","owner":[],"postedDate":"March 24th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-24T13:51:47+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-24 13:51:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9019468","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9019468","identity":"rs-9019468","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}