Medical Image Generation using Denoising Diffusion Probabilistic Model

Medical Image Generation using Denoising Diffusion Probabilistic Model | 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 Medical Image Generation using Denoising Diffusion Probabilistic Model Saritha A N, Sarvesh Rastogi, Shreya Bharamanna Patil, Basavaraj Talawar, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8212171/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 Denoising Diffusion Probabilistic Models (DDPMs) have emerged as a new state of the art for image generation. Unlike previous approaches, like GANs and VAEs (which sometimes have difficulty during training or suffer from issues such as mode collapse), DDPMs follow a more stable approach that gradually diminishes noise in an image over many steps. While this method is less efficient, it tends to yield cleaner and more realistic results. In this work, we propose a class-conditional DDPM for synthesizing skin lesion images with a curated Skin Diseases Dataset. The image data consists of three lesion categories: benign keratosis-like lesions (BKL), melanocytic nevi (NV) and vascular lesions (VASC). Our model leverages a U-Net to predict the noise added by time-point in diffusion and takes class input to enforce that the generated images belong to the desired lesion category. Training is performed with the usual DDPM objective, based on mean squared error between predicted noise and actual noise under a fixed noise-averaging schedule. The results demonstrate that our model can effectively synthesized images for all the three classes, and simulated appearance-alike visual characteristics fit to the evolutionary habit and manifestation of various types of lesions. To investigate how useful these synthetic images are for various data generation beyond s generation, we also analyze their effect on a downstream classification task. We form several variants of real images and DDPM-generated ones and train another classifier to investigate whether this type of augmentation can enhance diagnosis. In this work, we show both the generative capabilities of our model as well as the benefits of introducing synthetic data to medical image classification workflows. stable diffusion models DDPM U-NET medical image medical image augmentation YOLOv8 Skin diseases FID score deep learning GANs medical image generation HAM10000 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-8212171","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":571544847,"identity":"2b2a4413-378d-42db-8813-a7b14a32c52f","order_by":0,"name":"Saritha A N","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Saritha","middleName":"A","lastName":"N","suffix":""},{"id":571544848,"identity":"672ae944-fe87-48f6-a964-24704dd6d593","order_by":1,"name":"Sarvesh Rastogi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIiWNgGAWjYPCCA0DMfPDBxwa4iAQxWtiSDWeSqIXHTJq3gYA6EJBvP/tM4sOfOwz8/QcMpG13bLPnb29g/PCDwSIPlxaDM+lmkjPbnjFI3EhIMM49cztxxpkDzJI9DBLFOLUwpDEb8zYcZjCQYDiQnNt2O8FAIoFBGuiXRFyOlO9/xmz85w9QC//BhsOWbbftDeQfMP/Gp4XhRhrjYwY2oBaGZMZmxrbbjBskGNjw2mJw4xnjw962wzwSN9KYGXvBfklss+wxwOewNIYDP/4cluPvP//9x88dt4EhdvjwjR8VdbgdBgU8SGzGBlCwjIJRMApGwSigAAAAWlBYo+CiCYIAAAAASUVORK5CYII=","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":true,"prefix":"","firstName":"Sarvesh","middleName":"","lastName":"Rastogi","suffix":""},{"id":571544849,"identity":"b3e322ce-a8a1-4121-9e9f-fd377d24c9f0","order_by":2,"name":"Shreya Bharamanna Patil","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Shreya","middleName":"Bharamanna","lastName":"Patil","suffix":""},{"id":571544850,"identity":"8cfe8e17-3ac1-4314-9808-34c269cd60e0","order_by":3,"name":"Basavaraj Talawar","email":"","orcid":"","institution":"National Institute of Technology Karnataka","correspondingAuthor":false,"prefix":"","firstName":"Basavaraj","middleName":"","lastName":"Talawar","suffix":""},{"id":571544852,"identity":"2dbcc80d-0ea6-4fef-a6e1-81cd6cdaebcb","order_by":4,"name":"Shreya Soni","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Shreya","middleName":"","lastName":"Soni","suffix":""},{"id":571544854,"identity":"f4e4f52e-75bf-4ac1-b0df-7aca2dbaed29","order_by":5,"name":"Setu Mishra","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Setu","middleName":"","lastName":"Mishra","suffix":""}],"badges":[],"createdAt":"2025-11-26 11:08:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8212171/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8212171/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99995162,"identity":"00aab4d6-16af-4783-a7a6-5dc2d0344197","added_by":"auto","created_at":"2026-01-12 02:48:47","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7388,"visible":true,"origin":"","legend":"","description":"","filename":"1b583492db2947e9acef46fd8d44803e.json","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/4a14254de032465b6339f94a.json"},{"id":100360191,"identity":"4ef0faab-6ee7-4197-b716-badcefc30bb8","added_by":"auto","created_at":"2026-01-16 07:37:56","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":123782,"visible":true,"origin":"","legend":"","description":"","filename":"Comparison.png","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/7863e91ced29630c3d5e9e3a.png"},{"id":99995163,"identity":"f7001a54-89ac-4197-bf9b-2905e6e0c9d1","added_by":"auto","created_at":"2026-01-12 02:48:47","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":519323,"visible":true,"origin":"","legend":"","description":"","filename":"FinalDDPMamended.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/3c1acc28adafde130906e739.pdf"},{"id":99995171,"identity":"a4c30b2d-5d69-4742-9be5-795606cda62b","added_by":"auto","created_at":"2026-01-12 02:48:48","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":519323,"visible":true,"origin":"","legend":"","description":"","filename":"FinalDDPMamendedd1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/6174b63eee2d32cc1e5bd8c0.pdf"},{"id":99995166,"identity":"7950ff7f-1add-48d8-9a2c-3b42bf69fcc0","added_by":"auto","created_at":"2026-01-12 02:48:47","extension":"bst","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":61107,"visible":true,"origin":"","legend":"","description":"","filename":"bmcmathphys.bst","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/d3268e50ab2a4a43c4991280.bst"},{"id":99995164,"identity":"8329522e-a230-4ef9-a7e0-665b2c8be4ce","added_by":"auto","created_at":"2026-01-12 02:48:47","extension":"sty","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11595,"visible":true,"origin":"","legend":"","description":"","filename":"bmcartbiblio.sty","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/2f3ab5ea97d4d1abff1fae02.sty"},{"id":100360869,"identity":"90652ed6-d097-4a1c-9bbc-682d46940892","added_by":"auto","created_at":"2026-01-16 07:44:04","extension":"cls","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55703,"visible":true,"origin":"","legend":"","description":"","filename":"bmcart.cls","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/241c8b0789f98e4d194dd2cb.cls"},{"id":99995173,"identity":"22992cfc-1559-4621-a845-b15b65b65c7c","added_by":"auto","created_at":"2026-01-12 02:48:48","extension":"jpg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38088,"visible":true,"origin":"","legend":"","description":"","filename":"fidplot.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/a7cbbd5f068257bf279dcd02.jpg"},{"id":99995168,"identity":"4887c7bd-957c-4bbf-8253-687097a6dd39","added_by":"auto","created_at":"2026-01-12 02:48:48","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21064,"visible":true,"origin":"","legend":"","description":"","filename":"results.png","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/e731f4d7021a90e93ebb1d53.png"},{"id":100360174,"identity":"1a7eb026-0f12-405e-b035-b9a226b23342","added_by":"auto","created_at":"2026-01-16 07:37:48","extension":"bst","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33250,"visible":true,"origin":"","legend":"","description":"","filename":"spbasic.bst","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/9aa913772385d3551568187f.bst"},{"id":99995169,"identity":"ab1ecdd6-98a2-40c3-a1bf-584db23e3fbd","added_by":"auto","created_at":"2026-01-12 02:48:48","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14184,"visible":true,"origin":"","legend":"","description":"","filename":"updatedmodel.png","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/e4a0ac56b18ec3f27d839d4c.png"},{"id":100360213,"identity":"18bfa96c-9642-4fff-a556-97f7143dbd46","added_by":"auto","created_at":"2026-01-16 07:38:02","extension":"bst","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38215,"visible":true,"origin":"","legend":"","description":"","filename":"vancouver.bst","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/29eaa0cc566d54a80fae45b4.bst"},{"id":100360094,"identity":"4c8f4c1c-d470-4a91-aec1-087c8535fb89","added_by":"auto","created_at":"2026-01-16 07:36:59","extension":"xml","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":69221,"visible":true,"origin":"","legend":"","description":"","filename":"1b583492db2947e9acef46fd8d44803e1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1/378ecb41a9f9bd6e0b1f5a24.xml"},{"id":107043958,"identity":"51ce09d6-4809-4159-8d61-39ea70fa2424","added_by":"auto","created_at":"2026-04-16 06:57:45","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":413529,"visible":true,"origin":"","legend":"","description":"","filename":"FinalDDPMamended.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8212171/v1_covered_6f73fc2a-c7eb-4633-b4cf-87975868f4a5.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Medical Image Generation using Denoising Diffusion Probabilistic Model ","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":"stable diffusion models, DDPM, U-NET, medical image, medical image augmentation, YOLOv8, Skin diseases, FID score, deep learning, GANs, medical image generation, HAM10000","lastPublishedDoi":"10.21203/rs.3.rs-8212171/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8212171/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Denoising Diffusion Probabilistic Models (DDPMs) have emerged as a new state of the art for image generation. Unlike previous approaches, like GANs and VAEs (which sometimes have difficulty during training or suffer from issues such as mode collapse), DDPMs follow a more stable approach that gradually diminishes noise in an image over many steps. While this method is less efficient, it tends to yield cleaner and more realistic results. In this work, we propose a class-conditional DDPM for synthesizing skin lesion images with a curated Skin Diseases Dataset. The image data consists of three lesion categories: benign keratosis-like lesions (BKL), melanocytic nevi (NV) and vascular lesions (VASC). Our model leverages a U-Net to predict the noise added by time-point in diffusion and takes class input to enforce that the generated images belong to the desired lesion category. Training is performed with the usual DDPM objective, based on mean squared error between predicted noise and actual noise under a fixed noise-averaging schedule. The results demonstrate that our model can effectively synthesized images for all the three classes, and simulated appearance-alike visual characteristics fit to the evolutionary habit and manifestation of various types of lesions. To investigate how useful these synthetic images are for various data generation beyond s generation, we also analyze their effect on a downstream classification task. We form several variants of real images and DDPM-generated ones and train another classifier to investigate whether this type of augmentation can enhance diagnosis. In this work, we show both the generative capabilities of our model as well as the benefits of introducing synthetic data to medical image classification workflows.","manuscriptTitle":"Medical Image Generation using Denoising Diffusion Probabilistic Model","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-12 02:48:43","doi":"10.21203/rs.3.rs-8212171/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":"e4c1b950-024c-4989-9894-0e6ead0a5470","owner":[],"postedDate":"January 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-16T06:53:48+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-12 02:48:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8212171","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8212171","identity":"rs-8212171","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}