{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,3]],"date-time":"2026-05-03T04:51:31Z","timestamp":1777783891859,"version":"3.51.4"},"reference-count":35,"publisher":"BMJ","issue":"12","funder":[{"name":"Kim Ki-Soo Scholarship Committee","award":["2021"],"award-info":[{"award-number":["2021"]}]},{"DOI":"10.13039\/501100003725","name":"National Research Foundation of Korea","doi-asserted-by":"publisher","award":["NRF001663041G0003101"],"award-info":[{"award-number":["NRF001663041G0003101"]}],"id":[{"id":"10.13039\/501100003725","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["bmj.com"],"crossmark-restriction":true},"short-container-title":["Br J Ophthalmol"],"accepted":{"date-parts":[[2022,9,27]]},"published-print":{"date-parts":[[2023,12]]},"abstract":"<jats:sec>\n<jats:title>Background\/Aims<\/jats:title>\n<jats:p>Retinal capillary non-perfusion (NP) and neovascularisation (NV) are two of the most important angiographic changes in diabetic retinopathy (DR). This study investigated the feasibility of using deep learning (DL) models to automatically segment NP and NV on ultra-widefield fluorescein angiography (UWFA) images from patients with DR.<\/jats:p>\n<\/jats:sec>\n<jats:sec>\n<jats:title>Methods<\/jats:title>\n<jats:p>Retrospective cross-sectional chart review study. In total, 951 UWFA images were collected from patients with severe non-proliferative DR (NPDR) or proliferative DR (PDR). Each image was segmented and labelled for NP, NV, disc, background and outside areas. Using the labelled images, DL models were trained and validated (80%) using convolutional neural networks (CNNs) for automated segmentation and tested (20%) on test sets. Accuracy of each model and each label were assessed.<\/jats:p>\n<\/jats:sec>\n<jats:sec>\n<jats:title>Results<\/jats:title>\n<jats:p>The best accuracy from CNN models for each label was 0.8208, 0.8338, 0.9801, 0.9253 and 0.9766 for NP, NV, disc, background and outside areas, respectively. The best Intersection over Union for each label was 0.6806, 0.5675, 0.7107, 0.8551 and 0.924 and mean mean boundary F1 score (BF score) was 0.6702, 0.8742, 0.9092, 0.8103 and 0.9006, respectively.<\/jats:p>\n<\/jats:sec>\n<jats:sec>\n<jats:title>Conclusions<\/jats:title>\n<jats:p>DL models can detect NV and NP as well as disc and outer margins on UWFA with good performance. This automated segmentation of important UWFA features will aid physicians in DR clinics and in overcoming grader subjectivity.<\/jats:p>\n<\/jats:sec>","DOI":"10.1136\/bjo-2022-321063","type":"journal-article","created":{"date-parts":[[2022,10,14]],"date-time":"2022-10-14T17:27:17Z","timestamp":1665768437000},"page":"1859-1863","update-policy":"https:\/\/doi.org\/10.1136\/crossmarkpolicy","source":"Crossref","is-referenced-by-count":12,"title":["Automated segmentation of ultra-widefield fluorescein angiography of diabetic retinopathy using deep learning"],"prefix":"10.1136","volume":"107","author":[{"given":"Phil-Kyu","family":"Lee","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ho","family":"Ra","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6736-5379","authenticated-orcid":false,"given":"Jiwon","family":"Baek","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"239","published-online":{"date-parts":[[2022,10,14]]},"reference":[{"key":"2024110109351134000_107.12.1859.1","doi-asserted-by":"crossref","first-page":"16012","DOI":"10.1038\/nrdp.2016.12","article-title":"Diabetic retinopathy","volume":"2","author":"Wong","year":"2016","journal-title":"Nat Rev Dis Primers"},{"key":"2024110109351134000_107.12.1859.2","doi-asserted-by":"crossref","first-page":"2170","DOI":"10.1016\/j.ophtha.2009.04.043","article-title":"Blindness in a 25-year follow-up of a population-based cohort of danish type 1 diabetic patients","volume":"116","author":"Grauslund","year":"2009","journal-title":"Ophthalmology"},{"key":"2024110109351134000_107.12.1859.3","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(09)62124-3"},{"key":"2024110109351134000_107.12.1859.4","doi-asserted-by":"crossref","first-page":"2443","DOI":"10.1002\/jcb.25575","article-title":"Angiogenesis and inflammation crosstalk in diabetic retinopathy","volume":"117","author":"Capit\u00e3o","year":"2016","journal-title":"J Cell Biochem"},{"key":"2024110109351134000_107.12.1859.5","doi-asserted-by":"publisher","DOI":"10.1016\/j.ophtha.2009.06.051"},{"key":"2024110109351134000_107.12.1859.6","doi-asserted-by":"publisher","DOI":"10.2337\/diacare.27.10.2540"},{"key":"2024110109351134000_107.12.1859.7","doi-asserted-by":"crossref","first-page":"140","DOI":"10.1159\/000308758","article-title":"Retinal ischemia (capillary nonperfusion) and retinal neovascularization in patients with diabetic retinopathy","volume":"177","author":"Merin","year":"1978","journal-title":"Ophthalmologica"},{"key":"2024110109351134000_107.12.1859.8","doi-asserted-by":"crossref","first-page":"367","DOI":"10.1001\/jamaophthalmol.2015.5658","article-title":"Automated quantification of capillary nonperfusion using optical coherence tomography angiography in diabetic retinopathy","volume":"134","author":"Hwang","year":"2016","journal-title":"JAMA Ophthalmol"},{"key":"2024110109351134000_107.12.1859.9","doi-asserted-by":"crossref","first-page":"415","DOI":"10.1016\/j.oret.2019.10.018","article-title":"Ultra-widefield protocol enhances automated classification of diabetic retinopathy severity with OCT angiography","volume":"4","author":"Wang","year":"2020","journal-title":"Ophthalmol Retina"},{"key":"2024110109351134000_107.12.1859.10","doi-asserted-by":"publisher","DOI":"10.1016\/S0161-6420(81)34983-5"},{"key":"2024110109351134000_107.12.1859.11","doi-asserted-by":"publisher","DOI":"10.1016\/j.ophtha.2015.07.034"},{"key":"2024110109351134000_107.12.1859.12","doi-asserted-by":"publisher","DOI":"10.1001\/archopht.1987.01060060096041"},{"key":"2024110109351134000_107.12.1859.13","doi-asserted-by":"publisher","DOI":"10.1364\/BOE.9.005147"},{"key":"2024110109351134000_107.12.1859.14","doi-asserted-by":"publisher","DOI":"10.1016\/j.ophtha.2013.05.004"},{"key":"2024110109351134000_107.12.1859.15","doi-asserted-by":"publisher","DOI":"10.1097\/IAE.0b013e3182278b64"},{"key":"2024110109351134000_107.12.1859.16","doi-asserted-by":"publisher","DOI":"10.1136\/bjophthalmol-2011-300774"},{"key":"2024110109351134000_107.12.1859.17","first-page":"1300","article-title":"Automated detection of neovascularization for proliferative diabetic retinopathy screening","volume":"2016","author":"Roychowdhury","year":"2016","journal-title":"Annu Int Conf IEEE Eng Med Biol Soc"},{"key":"2024110109351134000_107.12.1859.18","doi-asserted-by":"publisher","DOI":"10.1007\/s10278-011-9418-6"},{"key":"2024110109351134000_107.12.1859.19","doi-asserted-by":"crossref","DOI":"10.3390\/s21165327","article-title":"Neovascularization detection and localization in fundus images using deep learning","volume":"21","author":"Tang","year":"2021","journal-title":"Sensors"},{"key":"2024110109351134000_107.12.1859.20","doi-asserted-by":"crossref","first-page":"1170","DOI":"10.1001\/jamaophthalmol.2017.3782","article-title":"Automated grading of age-related macular degeneration from color fundus images using deep convolutional neural networks","volume":"135","author":"Burlina","year":"2017","journal-title":"JAMA Ophthalmol"},{"key":"2024110109351134000_107.12.1859.21","doi-asserted-by":"publisher","DOI":"10.1016\/j.ophtha.2017.02.008"},{"key":"2024110109351134000_107.12.1859.22","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2017.18152"},{"key":"2024110109351134000_107.12.1859.23","doi-asserted-by":"publisher","DOI":"10.1016\/S0161-6420(13)38014-2"},{"key":"2024110109351134000_107.12.1859.24","doi-asserted-by":"publisher","DOI":"10.1007\/BF02150141"},{"key":"2024110109351134000_107.12.1859.25","doi-asserted-by":"publisher","DOI":"10.1136\/bjophthalmol-2018-313173"},{"key":"2024110109351134000_107.12.1859.26","doi-asserted-by":"crossref","DOI":"10.3390\/a12080154","article-title":"A survey of convolutional neural networks on edge with reconfigurable computing","volume":"12","author":"V\u00e9stias","year":"2019","journal-title":"Algorithms"},{"key":"2024110109351134000_107.12.1859.27","doi-asserted-by":"publisher","DOI":"10.1167\/iovs.16-19776"},{"key":"2024110109351134000_107.12.1859.28","doi-asserted-by":"crossref","first-page":"183","DOI":"10.15171\/bi.2015.27","article-title":"Detection of retinal capillary nonperfusion in fundus fluorescein angiogram of diabetic retinopathy","volume":"5","author":"Rasta","year":"2015","journal-title":"Bioimpacts"},{"key":"2024110109351134000_107.12.1859.29","doi-asserted-by":"crossref","first-page":"bjophthalmol-2018-313596","DOI":"10.1136\/bjophthalmol-2018-313596","article-title":"Analysis of quantitative correlations between microaneurysm, ischaemic index and new vessels in ultrawide-field fluorescein angiography images using automated software","volume":"103","author":"Son","year":"2019","journal-title":"Br J Ophthalmol"},{"key":"2024110109351134000_107.12.1859.30","doi-asserted-by":"crossref","DOI":"10.3390\/jcm9051462","article-title":"Relationships among retinal nonperfusion, neovascularization, and vascular endothelial growth factor levels in quiescent proliferative diabetic retinopathy","volume":"9","author":"Ra","year":"2020","journal-title":"J Clin Med"},{"key":"2024110109351134000_107.12.1859.31","doi-asserted-by":"publisher","DOI":"10.1167\/iovs.61.12.14"},{"key":"2024110109351134000_107.12.1859.32","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-020-71622-6","article-title":"Automatic detection of non-perfusion areas in diabetic macular edema from fundus fluorescein angiography for decision making using deep learning","volume":"10","author":"Jin","year":"2020","journal-title":"Sci Rep"},{"key":"2024110109351134000_107.12.1859.33","doi-asserted-by":"crossref","first-page":"648","DOI":"10.1002\/mp.14640","article-title":"Automated segmentation of retinal nonperfusion area in fluorescein angiography in retinal vein occlusion using convolutional neural networks","volume":"48","author":"Tang","year":"2021","journal-title":"Med Phys"},{"key":"2024110109351134000_107.12.1859.34","doi-asserted-by":"crossref","DOI":"10.3390\/jcm9082537","article-title":"Deep learning-based segmentation and quantification of retinal capillary non-perfusion on ultra-wide-field retinal fluorescein angiography","volume":"9","author":"Nunez do Rio","year":"2020","journal-title":"J Clin Med"},{"key":"2024110109351134000_107.12.1859.35","doi-asserted-by":"crossref","first-page":"346","DOI":"10.1016\/j.compmedimag.2013.06.008","article-title":"Detection of neovascularization in retinal images using multivariate m-mediods based classifier","volume":"37","author":"Usman Akram","year":"2013","journal-title":"Comput Med Imaging Graph"}],"container-title":["British Journal of Ophthalmology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1136\/bjo-2022-321063","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,4]],"date-time":"2024-11-04T17:00:41Z","timestamp":1730739641000},"score":1,"resource":{"primary":{"URL":"https:\/\/bjo.bmj.com\/lookup\/doi\/10.1136\/bjo-2022-321063"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,10,14]]},"references-count":35,"journal-issue":{"issue":"12","published-online":{"date-parts":[[2023,11,22]]},"published-print":{"date-parts":[[2023,12]]}},"alternative-id":["10.1136\/bjo-2022-321063"],"URL":"https:\/\/doi.org\/10.1136\/bjo-2022-321063","relation":{},"ISSN":["0007-1161","1468-2079"],"issn-type":[{"value":"0007-1161","type":"print"},{"value":"1468-2079","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,10,14]]}}}