{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T20:59:49Z","timestamp":1785185989745,"version":"3.55.0"},"reference-count":44,"publisher":"Wiley","license":[{"start":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T00:00:00Z","timestamp":1785110400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"},{"start":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T00:00:00Z","timestamp":1785110400000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Acta Ophthalmologica"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>To develop and validate a self\u2010supervised vision transformer (ViT) for automated three\u2010class classification of Normal, High Myopia (HM), and Pathologic Myopia (PM) from colour fundus photographs, with anatomically faithful interpretability.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      A DINOv2 self\u2010supervised ViT was fine\u2010tuned using a two\u2010stage transfer\u2010learning protocol with class\u2010balanced sampling. Model robustness was assessed using 10\u2010fold stratified cross\u2010validation, and an ensemble of the best fold\u2010specific checkpoints was evaluated once on a strictly held\u2010out independent test set. Comparative performance was assessed against ResNet\u201050, VGG\u201016, and EfficientNet\u2010B3 using paired, two\u2010sided Student's\n                      <jats:italic>t<\/jats:italic>\n                      \u2010tests on fold\u2010level macro\u2013F1 scores. Model explainability was examined using a Dual\u2010Rollout Attention approach that fuses early\u2010layer spatial detail with late\u2010layer semantic information.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      On the independent test set, the ensemble achieved an overall accuracy of 97.03% (95% CI 96.75\u201397.32%) and a macro F1\u2010score of 97.08% (95% CI 96.80\u201397.36%). The DINOv2\u2010based framework outperformed ResNet\u201050, VGG\u201016, and EfficientNet\u2010B3 architectures (all\n                      <jats:italic>p<\/jats:italic>\n                      \u2009&lt;\u20090.001). Dual\u2010Rollout Attention highlighted clinically relevant regions including optic disc morphology, peripapillary atrophy, and areas consistent with chorioretinal thinning.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>A self\u2010supervised ViT combined with transformer\u2010specific interpretability enables accurate, robust, and transparent automated classification of high and pathologic myopia from fundus photographs. This approach may support standardized assessment and scalable screening, pending external and prospective validation.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1111\/aos.70205","type":"journal-article","created":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T20:17:38Z","timestamp":1785183458000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Explainable deep learning self\u2010supervised vision transformer for fundus\u2010based myopia classification"],"prefix":"10.1111","author":[{"ORCID":"https:\/\/orcid.org\/0009-0008-3679-8807","authenticated-orcid":false,"given":"Eirini","family":"Maliagkani","sequence":"first","affiliation":[{"name":"1st Department of Ophthalmology General Hospital of Athens \u201cG.\u00a0Gennimatas\u201d, National and Kapodistrian University of Athens  Athens Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christos","family":"Tsoutsas","sequence":"additional","affiliation":[{"name":"ACTA Lab, Department of Energy Systems University of Thessaly  Larisa Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nikolaos","family":"Papageorgiou","sequence":"additional","affiliation":[{"name":"ACTA Lab, Department of Energy Systems University of Thessaly  Larisa Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6439-9282","authenticated-orcid":false,"given":"Ioannis D.","family":"Apostolopoulos","sequence":"additional","affiliation":[{"name":"ACTA Lab, Department of Energy Systems University of Thessaly  Larisa Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nikolaos","family":"Papandrianos","sequence":"additional","affiliation":[{"name":"ACTA Lab, Department of Energy Systems University of Thessaly  Larisa Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ilias","family":"Georgalas","sequence":"additional","affiliation":[{"name":"1st Department of Ophthalmology General Hospital of Athens \u201cG.\u00a0Gennimatas\u201d, National and Kapodistrian University of Athens  Athens Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Elpiniki","family":"Papageorgiou","sequence":"additional","affiliation":[{"name":"ACTA Lab, Department of Energy Systems University of Thessaly  Larisa Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,7,27]]},"reference":[{"key":"e_1_2_8_2_1","doi-asserted-by":"crossref","first-page":"4190","DOI":"10.18653\/v1\/2020.acl-main.385","volume-title":"Proceedings of the 58th annual meeting of the Association for Computational Linguistics","author":"Abnar S.","year":"2020"},{"key":"e_1_2_8_3_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41746\u2010018\u20100040\u20106"},{"key":"e_1_2_8_4_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2023.103000"},{"key":"e_1_2_8_5_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ajo.2024.12.022"},{"key":"e_1_2_8_6_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ejrad.2023.110786"},{"key":"e_1_2_8_7_1","doi-asserted-by":"publisher","DOI":"10.1111\/aos.16693"},{"key":"e_1_2_8_8_1","first-page":"782","volume-title":"Proceedings of the IEEE\/CVF conference on computer vision and pattern recognition (CVPR)","author":"Chefer H.","year":"2021"},{"key":"e_1_2_8_9_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591\u2010018\u20100107\u20106"},{"key":"e_1_2_8_10_1","doi-asserted-by":"crossref","first-page":"248","DOI":"10.1109\/CVPR.2009.5206848","volume-title":"2009 IEEE conference on computer vision and pattern recognition (CVPR)","author":"Deng J.","year":"2009"},{"key":"e_1_2_8_11_1","unstructured":"Dosovitskiy A. Beyer L. Kolesnikov A. Weissenborn D. Zhai X. Unterthiner T.et\u00a0al. (2020)An image is worth 16\u2009\u00d7\u200916 words: transformers for image recognition at scale.arXivPreprint.https:\/\/arxiv.org\/abs\/2010.11929"},{"key":"e_1_2_8_12_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.oret.2021.02.006"},{"key":"e_1_2_8_13_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2022.11.035"},{"key":"e_1_2_8_14_1","doi-asserted-by":"publisher","DOI":"10.1167\/iovs.18\u201025957"},{"key":"e_1_2_8_15_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ajo.2024.09.035"},{"key":"e_1_2_8_16_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256\u2010020\u201000257\u2010z"},{"key":"e_1_2_8_17_1","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2016.17216"},{"key":"e_1_2_8_18_1","first-page":"770","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR)","author":"He K.","year":"2016"},{"key":"e_1_2_8_19_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.xops.2025.100923"},{"key":"e_1_2_8_20_1","unstructured":"Huang S. Li Z. Lin B. Zhang S. Yi Q.&Wang L.(2023)HPMI: a retinal fundus image dataset for identification of high and pathological myopia based on deep learning.Figshare.https:\/\/doi.org\/10.6084\/m9.figshare.24800232.v1"},{"key":"e_1_2_8_21_1","doi-asserted-by":"publisher","DOI":"10.7759\/cureus.78685"},{"key":"e_1_2_8_22_1","doi-asserted-by":"crossref","unstructured":"Li N. Li T. Hu C. Wang K.&Kang H.(2021)A benchmark of ocular disease intelligent recognition: one shot for multi\u2010disease detection.arXiv.https:\/\/doi.org\/10.48550\/arXiv.2102.07978","DOI":"10.1007\/978-3-030-71058-3_11"},{"key":"e_1_2_8_23_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2017.07.005"},{"key":"e_1_2_8_24_1","unstructured":"Loshchilov I.&Hutter F.(2019)Decoupled weight decay regularization. In International conference on learning representations (ICLR). arXiv.https:\/\/arxiv.org\/abs\/1711.05101"},{"key":"e_1_2_8_25_1","doi-asserted-by":"publisher","DOI":"10.3390\/jimaging9040084"},{"key":"e_1_2_8_26_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ajo.2015.01.022"},{"key":"e_1_2_8_27_1","doi-asserted-by":"publisher","DOI":"10.1167\/iovs.62.5.5"},{"key":"e_1_2_8_28_1","unstructured":"Oquab M. Darcet T. Moutakanni T. Vo H. Szafraniec M. Khalidov V.et\u00a0al. (2024)DINOv2: learning robust visual features without supervision.arXiv.https:\/\/arxiv.org\/abs\/2304.07193"},{"key":"e_1_2_8_29_1","doi-asserted-by":"publisher","DOI":"10.1111\/aos.70001"},{"key":"e_1_2_8_30_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41598\u2010025\u201010548\u20103"},{"key":"e_1_2_8_31_1","first-page":"12116","article-title":"Do vision transformers see like convolutional neural networks?","volume":"34","author":"Raghu M.","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_2_8_32_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256\u2010019\u20100048\u2010x"},{"key":"e_1_2_8_33_1","doi-asserted-by":"publisher","DOI":"10.1167\/iovs.62.5.2"},{"key":"e_1_2_8_34_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2023.102802"},{"key":"e_1_2_8_35_1","unstructured":"Simonyan K.&Zisserman A.(2014)Very deep convolutional networks for large\u2010scale image recognition.arXiv.https:\/\/arxiv.org\/abs\/1409.1556"},{"key":"e_1_2_8_36_1","doi-asserted-by":"publisher","DOI":"10.5555\/2627435.2670313"},{"key":"e_1_2_8_37_1","first-page":"6105","volume-title":"Proceedings of the 36th international conference on machine learning (ICML)","author":"Tan M.","year":"2019"},{"key":"e_1_2_8_38_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.aci.2018.08.003"},{"key":"e_1_2_8_39_1","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2017.18152"},{"key":"e_1_2_8_40_1","doi-asserted-by":"publisher","DOI":"10.1136\/bjophthalmol\u20102018\u2010313173"},{"key":"e_1_2_8_41_1","first-page":"10347","volume-title":"Proceedings of the 38th international conference on machine learning (ICML)","author":"Touvron H.","year":"2021"},{"key":"e_1_2_8_42_1","volume-title":"PyTorch image models (timm) [software]","author":"Wightman R.","year":"2019"},{"key":"e_1_2_8_43_1","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2013.281053"},{"key":"e_1_2_8_44_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.patcog.2023.109347"},{"key":"e_1_2_8_45_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41586\u2010023\u201006555\u2010x"}],"container-title":["Acta Ophthalmologica"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/aos.70205","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1111\/aos.70205","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/aos.70205","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T20:17:42Z","timestamp":1785183462000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1111\/aos.70205"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,27]]},"references-count":44,"alternative-id":["10.1111\/aos.70205"],"URL":"https:\/\/doi.org\/10.1111\/aos.70205","archive":["Portico"],"relation":{},"ISSN":["1755-375X","1755-3768"],"issn-type":[{"value":"1755-375X","type":"print"},{"value":"1755-3768","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,27]]},"assertion":[{"value":"2026-01-26","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-07-01","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-07-27","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"aos.70205"}}