{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,22]],"date-time":"2026-09-22T17:59:28Z","timestamp":1790099968913,"version":"4.0.1"},"reference-count":22,"publisher":"Ovid Technologies (Wolters Kluwer Health)","issue":"10","funder":[{"name":"Natural Science Foundation of Jilin Province","award":["No. YDZJ202401360ZYTS"],"award-info":[{"award-number":["No. YDZJ202401360ZYTS"]}]}],"content-domain":{"domain":["lww.com","ovid.com"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,10]]},"abstract":"<jats:sec>\n                    <jats:title>Purpose:<\/jats:title>\n                    <jats:p>To determine whether strictly curated synthetic images can improve grading performance in diabetic retinopathy (DR) diagnosis within limited data regimes, and to compare the utility of representative generative adversarial network and diffusion architectures.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Method:<\/jats:title>\n                    <jats:p>StyleGAN3 and Medfusion models were pretrained on an auxiliary data set. To prevent mode collapse and severity downshift, synthetic images were curated through a domain-matched Oracle network using Latent Space Rejection Sampling and Class-Conditioned SDEdit Escalation. Efficacy was evaluated on an isolated downstream DR classification task (Indian DR Image Dataset) across three architectures (ConvNeXt, ResNet50, VGG16) using five random seeds.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results:<\/jats:title>\n                    <jats:p>Curated StyleGAN3 data set achieved better Fr\u00e9chet inception distance (FID) (24.25) and padded structural similarity index measure (0.9037) compared with Medfusion (58.70 and 0.8852), although Medfusion yielded a higher inception score (2.35). In the isolated downstream classification task, baseline ConvNeXt models trained solely on real Indian DR Image Dataset data achieved a quadratic weighted kappa of 0.6350 \u00b1 0.0135. Augmentation with curated StyleGAN3 improved the quadratic weighted kappa to 0.7185 \u00b1 0.0309, while Medfusion augmentation achieved 0.7695 \u00b1 0.0254 (an absolute improvement of 0.1345). Crucially, Medfusion augmentation successfully doubled the sensitivity (Recall) for proliferative DR from 0.2615 \u00b1 0.1595 to 0.5385 \u00b1 0.0942.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion:<\/jats:title>\n                    <jats:p>Data augmentation using strictly curated synthetic images successfully addresses data imbalance and scarcity, significantly improving the ordinal accuracy and minority-class sensitivity of DR diagnosis models. Furthermore, standard global image quality metrics like FID may not fully capture localized pathological utility; diffusion models with higher FID scores ultimately provided superior structural diversity and downstream clinical performance.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1097\/iae.0000000000004919","type":"journal-article","created":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T17:02:25Z","timestamp":1784653345000},"page":"1759-1765","update-policy":"https:\/\/doi.org\/10.1097\/lww.0000000000001000","source":"Crossref","is-referenced-by-count":1,"title":["ENHANCING THE PERFORMANCE OF DIABETIC RETINOPATHY GRADING MODELS USING SYNTHETIC IMAGES"],"prefix":"10.1097","volume":"46","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7756-1226","authenticated-orcid":false,"given":"Jiashi","family":"Zhao","sequence":"first","affiliation":[{"name":"Department of Computer Science and Technology, Changchun University of Science and Technology, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yiting","family":"Zhang","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Technology, Changchun University of Science and Technology, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yu","family":"Miao","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Technology, Changchun University of Science and Technology, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Weili","family":"Shi","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Technology, Changchun University of Science and Technology, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jianhua","family":"Liu","sequence":"additional","affiliation":[{"name":"The Second Hospital of Jilin University, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhengang","family":"Jiang","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Technology, Changchun University of Science and Technology, Changchun, China."}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"276","reference":[{"key":"R1-20260922","doi-asserted-by":"crossref","first-page":"107843","DOI":"10.1016\/j.diabres.2019.107843","article-title":"Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: results from the International Diabetes Federation Diabetes Atlas, 9th edition","volume":"157","author":"Saeedi","year":"2019","journal-title":"Diabetes Res Clin Pract"},{"key":"R2-20260922","doi-asserted-by":"crossref","first-page":"102115","DOI":"10.1016\/j.bspc.2020.102115","article-title":"A lightweight CNN for diabetic retinopathy classification from fundus images","volume":"62","author":"Gayathri","year":"2020","journal-title":"Biomed Signal Process Control"},{"key":"R3-20260922","doi-asserted-by":"crossref","first-page":"231","DOI":"10.1038\/s41597-023-02125-y","article-title":"A large-scale synthetic pathological dataset for deep learning-enabled segmentation of breast cancer","volume":"10","author":"Ding","year":"2023","journal-title":"Sci Data"},{"key":"R5-20260922","doi-asserted-by":"crossref","first-page":"e2311878121","DOI":"10.1073\/pnas.2311878121","article-title":"Explaining neural scaling laws","volume":"121","author":"Bahri","year":"2024","journal-title":"Proc Natl Acad Sci USA"},{"key":"R6-20260922","doi-asserted-by":"crossref","first-page":"81","DOI":"10.3390\/jimaging9040081","article-title":"Deep learning approaches for data augmentation in medical imaging: a review","volume":"9","author":"Kebaili","year":"2023","journal-title":"J Imaging"},{"key":"R7-20260922","doi-asserted-by":"crossref","first-page":"100340","DOI":"10.1016\/j.health.2024.100340","article-title":"A systematic review of deep learning data augmentation in medical imaging: recent advances and future research directions","volume":"5","author":"Islam","year":"2024","journal-title":"Healthc Anal"},{"key":"R8-20260922","doi-asserted-by":"crossref","first-page":"368","DOI":"10.1038\/s43856-025-01092-2","article-title":"Forecasting the diabetic retinopathy progression using generative adversarial networks","volume":"5","author":"Qiao","year":"2025","journal-title":"Commun Med"},{"key":"R9-20260922","doi-asserted-by":"crossref","first-page":"131227","DOI":"10.1016\/j.eswa.2026.131227","article-title":"DR-DDPM: synthetic image generation for diabetic retinopathy using denoising diffusion probabilistic models for enhanced detection","volume":"309","author":"Shashank","year":"2026","journal-title":"Expert Syst Appl"},{"key":"R10-20260922","doi-asserted-by":"crossref","first-page":"791","DOI":"10.1136\/bjo-2024-326122","article-title":"Realistic fundus photograph generation for improving automated disease classification","volume":"109","author":"Pandey","year":"2025","journal-title":"Br J Ophthalmol"},{"key":"R11-20260922","doi-asserted-by":"crossref","first-page":"57","DOI":"10.1038\/s43856-025-01316-5","article-title":"Augmenting a prognostic deep learning system for referable diabetic retinopathy and maculopathy with synthetic retinal images","volume":"6","author":"Nderitu","year":"2025","journal-title":"Commun Med"},{"key":"R12-20260922","doi-asserted-by":"crossref","first-page":"25","DOI":"10.3390\/data3030025","article-title":"Indian diabetic retinopathy image dataset (IDRiD): a database for diabetic retinopathy screening research","volume":"3","author":"Porwal","year":"2018","journal-title":"Data"},{"key":"R13-20260922","first-page":"852","article-title":"Alias-free generative adversarial networks","volume":"34","author":"Karras","year":"2021","journal-title":"Adv Neural Inf Process Syst"},{"key":"R14-20260922","first-page":"12863","article-title":"Stylespace analysis: disentangled controls for stylegan image generation","author":"Wu","year":"2021"},{"key":"R15-20260922","doi-asserted-by":"crossref","first-page":"12098","DOI":"10.1038\/s41598-023-39278-0","article-title":"A multimodal comparison of latent denoising diffusion probabilistic models and generative adversarial networks for medical image synthesis","volume":"13","author":"M\u00fcller-Franzes","year":"2023","journal-title":"Sci Rep"},{"key":"R16-20260922","first-page":"10684","article-title":"High-resolution image synthesis with latent diffusion models","author":"Rombach","year":"2022"},{"key":"R17-20260922","first-page":"6840","article-title":"Denoising diffusion probabilistic models","volume":"33","author":"Ho","year":"2020","journal-title":"Adv Neural Inf Process Syst"},{"key":"R18-20260922","first-page":"234","article-title":"U-net: convolutional networks for biomedical image segmentation","author":"Ronneberger","year":"2015"},{"key":"R19-20260922","article-title":"Denoising diffusion implicit models","author":"Song","year":"2021"},{"key":"R20-20260922","first-page":"6629","article-title":"GANs trained by a two time-scale update rule converge to a local nash equilibrium","volume":"30","author":"Heusel","year":"2017","journal-title":"Adv Neural Inf Process Syst"},{"key":"R21-20260922","first-page":"2234","article-title":"Improved techniques for training GANs","author":"Salimans","year":"2016"},{"key":"R22-20260922","doi-asserted-by":"crossref","first-page":"600","DOI":"10.1109\/TIP.2003.819861","article-title":"Image quality assessment: from error visibility to structural similarity","volume":"13","author":"Wang","year":"2004","journal-title":"IEEE Trans Image Process"},{"key":"R23-20260922","doi-asserted-by":"crossref","first-page":"130","DOI":"10.1038\/s44172-025-00450-1","article-title":"Scorecard for synthetic medical data evaluation","volume":"4","author":"Zamzmi","year":"2025","journal-title":"Commun Eng"}],"container-title":["Retina"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.ovid.com\/10.1097\/IAE.0000000000004919","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,22]],"date-time":"2026-09-22T17:00:30Z","timestamp":1790096430000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ovid.com\/10.1097\/IAE.0000000000004919"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10]]},"references-count":22,"journal-issue":{"issue":"10","published-print":{"date-parts":[[2026]]}},"URL":"https:\/\/doi.org\/10.1097\/iae.0000000000004919","relation":{},"ISSN":["0275-004X","1539-2864"],"issn-type":[{"value":"0275-004X","type":"print"},{"value":"1539-2864","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10]]}}}