{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T00:46:42Z","timestamp":1793234802672,"version":"4.5.0"},"reference-count":43,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2025,12,20]],"date-time":"2025-12-20T00:00:00Z","timestamp":1766188800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,1,23]],"date-time":"2026-01-23T00:00:00Z","timestamp":1769126400000},"content-version":"vor","delay-in-days":34,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Commun Med"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Labelled data scarcity and class imbalance are common deep learning system (DLS) development challenges. We investigated if synthetic retinal images from a conditional cascaded diffusion model (CCDM) improves prognostic DLS (pDLS) performance for 2-year incident referable diabetic retinopathy or maculopathy (rDR\/rM) prediction.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Macula images from 72,559 eyes (September 2013 to December 2019) from the UK South-East London Diabetic Eye Screening Programme (SEL-DESP) formed the development dataset, whilst 9,071 eyes were used for internal testing. Images from 2,842 eyes from Birmingham DESP were used for external testing. Prognostic DLS were augmented with \u00d71, \u00d72, and \u00d74 additional synthetic positive cases (pDLS-G) and compared to unaugmented (pDLS-N) and \u00d71 positive-case resampled pDLS (pDLS-R) using the Area-Under-the Receiver Operating Characteristic curve (AUROC).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Here we show that CCDM generate realistic synthetic retinal images that are comparable to real images and demonstrate the utility of synthetic retinal images in augmenting the development of a pDLS. The internal and external test AUROC for the pDLS are 0.827 (95% CI: 0.794\u20130.861) and 0.756 (0.680\u20130.831), respectively. Augmentation with \u00d72 additional synthetic positive cases (pDLS-G \u00d72) significantly improves the internal test AUROC to 0.845 (95% CI: 0.812\u20130.877,\n                      <jats:italic>p<\/jats:italic>\n                      \u2009=\u20090.044) but does not improve the external test AUROC 0.717 (0.633\u20130.828,\n                      <jats:italic>p<\/jats:italic>\n                      \u2009=\u20090.243). Resampling positive real cases alone does not improve pDLS-R performance.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Augmenting pDLS with synthetic retinal images significantly improves pDLS performance on internal testing but not external testing suggesting further research is required to enhance the generalisability of synthetic retinal image augmentation.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1038\/s43856-025-01316-5","type":"journal-article","created":{"date-parts":[[2025,12,20]],"date-time":"2025-12-20T01:33:12Z","timestamp":1766194392000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":3,"title":["Augmenting a prognostic deep learning system for referable diabetic retinopathy and maculopathy with synthetic retinal images"],"prefix":"10.1038","volume":"6","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-8571-0751","authenticated-orcid":false,"given":"Paul","family":"Nderitu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8268-5389","authenticated-orcid":false,"given":"Joan M.","family":"Nunez do Rio","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Laura","family":"Webster","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0407-4501","authenticated-orcid":false,"given":"Samantha S.","family":"Mann","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Hopkins","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9152-3194","authenticated-orcid":false,"given":"Christos","family":"Bergeles","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Timothy L.","family":"Jackson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2025,12,20]]},"reference":[{"key":"1316_CR1","doi-asserted-by":"publisher","DOI":"10.1186\/s40537-019-0192-5","volume":"6","author":"JM Johnson","year":"2019","unstructured":"Johnson, J. M. & Khoshgoftaar, T. M. Survey on deep learning with class imbalance. J. Big Data 6, 27 (2019).","journal-title":"J. Big Data"},{"key":"1316_CR2","doi-asserted-by":"publisher","DOI":"10.1016\/j.xops.2022.100258","volume":"3","author":"YA Veturi","year":"2023","unstructured":"Veturi, Y. A. et al. SynthEye: investigating the impact of synthetic data on artificial intelligence-assisted gene diagnosis of inherited retinal disease. Ophthalmol. Sci. 3, 100258 (2023).","journal-title":"Ophthalmol. Sci."},{"key":"1316_CR3","unstructured":"Pinaya, W. H. L. et al. Brain Imaging Generation with Latent Diffusion Models 2022. https:\/\/ui.adsabs.harvard.edu\/abs\/2022arXiv220907162P (accessed 01 September 2022)."},{"key":"1316_CR4","unstructured":"Kingma, D. P. & Welling, M. Auto-Encoding Variational Bayes2013. https:\/\/ui.adsabs.harvard.edu\/abs\/2013arXiv1312.6114K (accessed 01 December 2013)."},{"key":"1316_CR5","unstructured":"van den Oord, A., Kalchbrenner, N. & Kavukcuoglu, K. Pixel Recurrent Neural Networks2016. https:\/\/ui.adsabs.harvard.edu\/abs\/2016arXiv160106759V (accessed 01 January 2016)."},{"key":"1316_CR6","doi-asserted-by":"publisher","unstructured":"Goodfellow, I. J. et al. Generative adversarial networks. Preprint at https:\/\/doi.org\/10.48550\/arXiv.1406.2661 (2014).","DOI":"10.48550\/arXiv.1406.2661"},{"key":"1316_CR7","doi-asserted-by":"publisher","unstructured":"Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2006.11239 (2020).","DOI":"10.48550\/arXiv.2006.11239"},{"key":"1316_CR8","doi-asserted-by":"publisher","unstructured":"Rombach, R., Blattmann, A., Lorenz, D., Esser, P. & Ommer, B. High-resolution image synthesis with latent diffusion models. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2112.10752 (2021).","DOI":"10.48550\/arXiv.2112.10752"},{"key":"1316_CR9","unstructured":"Dhariwal, P. & Nichol, A. Diffusion Models Beat GANs on Image Synthesis2021. https:\/\/ui.adsabs.harvard.edu\/abs\/2021arXiv210505233D (accessed 01 May 2021)."},{"key":"1316_CR10","doi-asserted-by":"publisher","unstructured":"Saharia, C. et al. Photorealistic text-to-image diffusion models with deep language understanding. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2205.11487 (2022).","DOI":"10.48550\/arXiv.2205.11487"},{"key":"1316_CR11","doi-asserted-by":"publisher","first-page":"56","DOI":"10.1109\/JBHI.2020.3045475","volume":"26","author":"Y Zhou","year":"2022","unstructured":"Zhou, Y., Wang, B., He, X., Cui, S. & Shao, L. DR-GAN: conditional generative adversarial network for fine-grained lesion synthesis on diabetic retinopathy images. IEEE J. Biomed. Health Inform. 26, 56\u201366 (2022).","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"1316_CR12","doi-asserted-by":"publisher","first-page":"533","DOI":"10.1364\/BOE.477906","volume":"14","author":"B Hou","year":"2023","unstructured":"Hou, B. High-fidelity diabetic retina fundus image synthesis from freestyle lesion maps. Biomed. Opt. Express 14, 533\u2013549 (2023).","journal-title":"Biomed. Opt. Express"},{"key":"1316_CR13","doi-asserted-by":"publisher","first-page":"2815","DOI":"10.1016\/j.bpj.2024.02.019","volume":"123","author":"J Liu","year":"2024","unstructured":"Liu, J. et al. VSG-GAN: a high-fidelity image synthesis method with semantic manipulation in retinal fundus image. Biophys. J. 123, 2815\u20132829 (2024).","journal-title":"Biophys. J."},{"key":"1316_CR14","doi-asserted-by":"crossref","unstructured":"Lim, G., Thombre, P., Lee, M. L. & Hsu, W. Generative data augmentation for diabetic retinopathy classification. In 2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI) 1096\u22121103 (IEEE, 2020)","DOI":"10.1109\/ICTAI50040.2020.00167"},{"key":"1316_CR15","doi-asserted-by":"crossref","unstructured":"Catapang, J. K., Santiago, D. E., Isip-Tan, I. T. Improving Detection of Diabetic Retinopathy in Low-Resolution Images via Latent Diffusion. In 2023 IEEE 6th International Conference on Multimedia Information Processing and Retrieval (MIPR) (IEEE, 2023)","DOI":"10.1109\/MIPR59079.2023.00024"},{"key":"1316_CR16","doi-asserted-by":"publisher","first-page":"167","DOI":"10.1038\/s43856-024-00590-z","volume":"4","author":"P Nderitu","year":"2024","unstructured":"Nderitu, P. et al. Predicting 1, 2 and 3 year emergent referable diabetic retinopathy and maculopathy using deep learning. Commun. Med. 4, 167 (2024).","journal-title":"Commun. Med."},{"key":"1316_CR17","doi-asserted-by":"publisher","first-page":"584","DOI":"10.1038\/s41591-023-02702-z","volume":"30","author":"L Dai","year":"2024","unstructured":"Dai, L. et al. A deep learning system for predicting time to progression of diabetic retinopathy. Nat. Med. 30, 584\u2013594 (2024).","journal-title":"Nat. Med."},{"key":"1316_CR18","doi-asserted-by":"publisher","first-page":"e10","DOI":"10.1016\/S2589-7500(20)30250-8","volume":"3","author":"A Bora","year":"2021","unstructured":"Bora, A. et al. Predicting the risk of developing diabetic retinopathy using deep learning.  Lancet Digital Health 3, e10\u2013e19 (2021).","journal-title":"Lancet Digital Health"},{"key":"1316_CR19","doi-asserted-by":"publisher","first-page":"e001140","DOI":"10.1136\/bmjophth-2022-001140","volume":"7","author":"Y Rom","year":"2022","unstructured":"Rom, Y., Aviv, R., Ianchulev, T. & Dvey-Aharon, Z. Predicting the future development of diabetic retinopathy using a deep learning algorithm for the analysis of non-invasive retinal imaging. BMJ Open Ophthalmol. 7, e001140 (2022).","journal-title":"BMJ Open Ophthalmol."},{"key":"1316_CR20","doi-asserted-by":"publisher","first-page":"833","DOI":"10.1136\/bjo-2023-323400","volume":"108","author":"J Mellor","year":"2024","unstructured":"Mellor, J. et al. Prediction of retinopathy progression using deep learning on retinal images within the Scottish screening programme. Br. J. Ophthalmol. 108, 833\u2013839 (2024).","journal-title":"Br. J. Ophthalmol."},{"key":"1316_CR21","unstructured":"Karras, T., Aittala, M., Aila, T. & Laine, S. Elucidating the design space of diffusion-based generative models 2022. https:\/\/ui.adsabs.harvard.edu\/abs\/2022arXiv220600364K (accessed 01 June 2022)."},{"key":"1316_CR22","unstructured":"Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B. & Hochreiter, S. Gans trained by a two time-scale update rule converge to a local Nash equilibrium. Adv. Neural Inform. Process. Syst. 30. (2017)."},{"key":"1316_CR23","doi-asserted-by":"publisher","first-page":"156","DOI":"10.1038\/s41586-023-06555-x","volume":"622","author":"Y Zhou","year":"2023","unstructured":"Zhou, Y. et al. A foundation model for generalizable disease detection from retinal images. Nature 622, 156\u2013163 (2023).","journal-title":"Nature"},{"key":"1316_CR24","doi-asserted-by":"publisher","first-page":"837","DOI":"10.2307\/2531595","volume":"44","author":"ER DeLong","year":"1988","unstructured":"DeLong, E. R., DeLong, D. M. & Clarke-Pearson, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44, 837\u2013845 (1988).","journal-title":"Biometrics"},{"key":"1316_CR25","doi-asserted-by":"publisher","first-page":"1191","DOI":"10.1016\/S0161-6420(85)33877-0","volume":"92","author":"BE Klein","year":"1985","unstructured":"Klein, B. E., Klein, R. & Moss, S. E. Prevalence of cataracts in a population-based study of persons with diabetes mellitus. Ophthalmology 92, 1191\u20131196 (1985).","journal-title":"Ophthalmology"},{"key":"1316_CR26","doi-asserted-by":"crossref","unstructured":"Rajesh, A. E. et al. Ethnicity is not biology: retinal pigment score to evaluate biological variability from ophthalmic imaging using machine learning. Nat Commun. 16, 60 (2023).","DOI":"10.1101\/2023.06.28.23291873"},{"key":"1316_CR27","doi-asserted-by":"publisher","first-page":"188","DOI":"10.1007\/s42452-024-05871-9","volume":"6","author":"HK Kim","year":"2024","unstructured":"Kim, H. K., Ryu, I. H., Choi, J. Y. & Yoo, T. K. A feasibility study on the adoption of a generative denoising diffusion model for the synthesis of fundus photographs using a small dataset. Discov. Appl. Sci. 6, 188 (2024).","journal-title":"Discov. Appl. Sci."},{"key":"1316_CR28","doi-asserted-by":"crossref","unstructured":"Go, S., Ji, Y., Park, S. J. & Lee, S. Generation of structurally realistic retinal fundus images with diffusion models. In Proc. IEEE\/CVF Conference on Computer Vision and Pattern Recognition; 2024. p. 2335-2344. (IEEE, 2024)","DOI":"10.1109\/CVPRW63382.2024.00239"},{"key":"1316_CR29","doi-asserted-by":"crossref","unstructured":"Alimanov A. & Islam M. B. Denoising diffusion probabilistic model for retinal image generation and segmentation. In 2023 IEEE international conference on computational photography (ICCP); IEEE. (IEEE, 2023).","DOI":"10.1109\/ICCP56744.2023.10233841"},{"key":"1316_CR30","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-022-20698-3","volume":"12","author":"M Kim","year":"2022","unstructured":"Kim, M. et al. Synthesizing realistic high-resolution retina image by style-based generative adversarial network and its utilization. Sci. Rep. 12, 17307 (2022).","journal-title":"Sci. Rep."},{"key":"1316_CR31","doi-asserted-by":"publisher","first-page":"100288","DOI":"10.1016\/j.chbr.2023.100288","volume":"10","author":"AF Di Natale","year":"2023","unstructured":"Di Natale, A. F., Simonetti, M. E., La Rocca, S. & Bricolo, E. Uncanny valley effect: a qualitative synthesis of empirical research to assess the suitability of using virtual faces in psychological research. Comput. Hum. Behav. Rep. 10, 100288 (2023).","journal-title":"Comput. Hum. Behav. Rep."},{"key":"1316_CR32","doi-asserted-by":"publisher","unstructured":"Ilanchezian, I. et al. Generating realistic counterfactuals for retinal fundus and OCT Images using Diffusion Models. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2311.11629 (2023).","DOI":"10.48550\/arXiv.2311.11629"},{"key":"1316_CR33","doi-asserted-by":"publisher","first-page":"60","DOI":"10.3390\/electronics11010060","volume":"11","author":"P Andreini","year":"2022","unstructured":"Andreini, P. et al. A two-stage GAN for high-resolution retinal image generation and segmentation. Electronics 11, 60 (2022).","journal-title":"Electronics"},{"key":"1316_CR34","doi-asserted-by":"publisher","first-page":"258","DOI":"10.1001\/jamaophthalmol.2018.6156","volume":"137","author":"PM Burlina","year":"2019","unstructured":"Burlina, P. M., Joshi, N., Pacheco, K. D., Liu, T. Y. A. & Bressler, N. M. Assessment of deep generative models for high-resolution synthetic retinal image generation of age-related macular degeneration. JAMA Ophthalmol. 137, 258\u2013264 (2019).","journal-title":"JAMA Ophthalmol."},{"key":"1316_CR35","doi-asserted-by":"publisher","first-page":"1283","DOI":"10.1007\/s11042-023-17058-2","volume":"83","author":"AQ Saeed","year":"2024","unstructured":"Saeed, A. Q., Sheikh Abdullah, S. N. H., Che-Hamzah, J., Abdul Ghani, A. T. & Abu-ain, W. A. K. Synthesizing retinal images using end-to-end VAEs-GAN pipeline-based sharpening and varying layer. Multimed. Tools Appl. 83, 1283\u20131307 (2024).","journal-title":"Multimed. Tools Appl."},{"key":"1316_CR36","doi-asserted-by":"publisher","DOI":"10.1186\/s12938-019-0682-x","volume":"18","author":"Z Yu","year":"2019","unstructured":"Yu, Z. et al. Retinal image synthesis from multiple-landmarks input with generative adversarial networks. Biomed. Eng. OnLine 18, 62 (2019).","journal-title":"Biomed. Eng. OnLine"},{"key":"1316_CR37","doi-asserted-by":"publisher","DOI":"10.1016\/j.xops.2021.100079","volume":"1","author":"JS Chen","year":"2021","unstructured":"Chen, J. S. et al. Deepfakes in ophthalmology: applications and realism of synthetic retinal images from generative adversarial networks. Ophthalmol. Sci. 1, 100079 (2021).","journal-title":"Ophthalmol. Sci."},{"key":"1316_CR38","first-page":"329","volume":"2020","author":"AS Coyner","year":"2020","unstructured":"Coyner, A. S. et al. Diagnosability of synthetic retinal fundus images for plus disease detection in retinopathy of prematurity. AMIA Annu. Symp. Proc. 2020, 329\u2013337 (2020).","journal-title":"AMIA Annu. Symp. Proc."},{"key":"1316_CR39","doi-asserted-by":"publisher","first-page":"3812865","DOI":"10.1155\/2021\/3812865","volume":"2021","author":"Y Chen","year":"2021","unstructured":"Chen, Y., Long, J. & Guo, J. RF-GANs: a method to synthesize retinal fundus images based on generative adversarial network. Comput. Intell. Neurosci. 2021, 3812865 (2021).","journal-title":"Comput. Intell. Neurosci."},{"key":"1316_CR40","first-page":"3855","volume":"63","author":"MJ Menten","year":"2022","unstructured":"Menten, M. J. et al. Discovery of imaging biomarkers for healthy aging and age-related macular degeneration using counterfactual generative adversarial networks. Investig. Ophthalmol. Vis. Sci. 63, 3855 (2022).","journal-title":"Investig. Ophthalmol. Vis. Sci."},{"key":"1316_CR41","doi-asserted-by":"publisher","first-page":"AIoa2300221","DOI":"10.1056\/AIoa2300221","volume":"1","author":"J Qiu","year":"2024","unstructured":"Qiu, J. et al. Development and validation of a multimodal multitask vision foundation model for generalist ophthalmic artificial intelligence. NEJM AI 1, AIoa2300221 (2024).","journal-title":"NEJM AI"},{"key":"1316_CR42","unstructured":"Radford, A. et al. Learning transferable visual models from natural language Supervision2021. https:\/\/ui.adsabs.harvard.edu\/abs\/2021arXiv210300020R (accessed 01 February, 2021)."},{"key":"1316_CR43","doi-asserted-by":"publisher","unstructured":"Paul, N. CCDM-Source-Code. 1.0 ed: Zenodo. https:\/\/doi.org\/10.5281\/zenodo.17761333. (2025).","DOI":"10.5281\/zenodo.17761333"}],"container-title":["Communications Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s43856-025-01316-5","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s43856-025-01316-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s43856-025-01316-5.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,23]],"date-time":"2026-01-23T14:45:12Z","timestamp":1769179512000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s43856-025-01316-5"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,20]]},"references-count":43,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["1316"],"URL":"https:\/\/doi.org\/10.1038\/s43856-025-01316-5","relation":{},"ISSN":["2730-664X"],"issn-type":[{"value":"2730-664X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,20]]},"assertion":[{"value":"27 June 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"5 December 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"20 December 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"P.N., J.M.N., L.W., D.H., and C.B. have no conflicts of interest to declare. S.S.M. has received speaker and advisory board fees from Bayer and Allergan and research grants from Novartis. S.S.M. has assisted in a study validating the EyeArt v2.1 DLS for diabetic retinopathy detection. T.J. is an advisor to 2CTech, Alcon, Bayer, Dutch Ophthalmic Research Centre, iLumen, Opthea, Outlook Therapeutics, Oxurion, Pfizer, and Regeneron, has received conference support from Roche, and free equipment use for non-commercial trials from Zeiss, Oraya, and LKC. T.L.J., P.N., and D.H. employer, King\u2019s College Hospital, has received funding for participants enroled to multiple commercial clinical trials including for DR.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"57"}}