Browse by author
Lookup NU author(s): Professor David SteelORCiD
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
© 2023, The Author(s). Medical artificial intelligence (AI) offers great potential for recognizing signs of health conditions in retinal images and expediting the diagnosis of eye diseases and systemic disorders 1. However, the development of AI models requires substantial annotation and models are usually task-specific with limited generalizability to different clinical applications 2. Here, we present RETFound, a foundation model for retinal images that learns generalizable representations from unlabelled retinal images and provides a basis for label-efficient model adaptation in several applications. Specifically, RETFound is trained on 1.6 million unlabelled retinal images by means of self-supervised learning and then adapted to disease detection tasks with explicit labels. We show that adapted RETFound consistently outperforms several comparison models in the diagnosis and prognosis of sight-threatening eye diseases, as well as incident prediction of complex systemic disorders such as heart failure and myocardial infarction with fewer labelled data. RETFound provides a generalizable solution to improve model performance and alleviate the annotation workload of experts to enable broad clinical AI applications from retinal imaging.
Author(s): Zhou Y, Chia MA, Wagner SK, Ayhan MS, Williamson DJ, Struyven RR, Liu T, Xu M, Lozano MG, Woodward-Court P, Kihara Y, Allen N, Gallacher JEJ, Littlejohns T, Aslam T, Bishop P, Black G, Sergouniotis P, Atan D, Dick AD, Williams C, Barman S, Barrett JH, Mackie S, Braithwaite T, Carare RO, Ennis S, Gibson J, Lotery AJ, Self J, Chakravarthy U, Hogg RE, Paterson E, Woodside J, Peto T, Mckay G, Mcguinness B, Foster PJ, Balaskas K, Khawaja AP, Pontikos N, Rahi JS, Lascaratos G, Patel PJ, Chan M, Chua SYL, Day A, Desai P, Egan C, Fruttiger M, Garway-Heath DF, Hardcastle A, Khaw SPT, Moore T, Sivaprasad S, Strouthidis N, Thomas D, Tufail A, Viswanathan AC, Dhillon B, Macgillivray T, Sudlow C, Vitart V, Doney A, Trucco E, Guggeinheim JA, Morgan JE, Hammond CJ, Williams K, Hysi P, Harding SP, Zheng Y, Luben R, Luthert P, Sun Z, McKibbin M, O'Sullivan E, Oram R, Weedon M, Owen CG, Rudnicka AR, Sattar N, Steel D, Stratton I, Tapp R, Yates MM, Petzold A, Madhusudhan S, Altmann A, Lee AY, Topol EJ, Denniston AK, Alexander DC, Keane PA
Publication type: Article
Publication status: Published
Journal: Nature
Year: 2023
Volume: 622
Pages: 156-163
Online publication date: 13/09/2023
Acceptance date: 18/08/2023
Date deposited: 27/09/2023
ISSN (print): 0028-0836
ISSN (electronic): 1476-4687
Publisher: Nature Publishing Group
URL: https://doi.org/10.1038/s41586-023-06555-x
DOI: 10.1038/s41586-023-06555-x
Data Access Statement: The MIDAS dataset consists of routinely collected healthcare data. Owing to its sensitive nature and the risk of reidentification, the dataset is subject to controlled access by means of a structured application process. Data access enquiries may be made to enquiries@insight.hdrhub.org and we will aim to respond within 2 weeks. Further details about the data request pipeline may be found on the INSIGHT Health Data Research Hub website https://www.insight.hdrhub.org. The AlzEye dataset is subject to the contractual restrictions of the data sharing agreements between National Health Service Digital, Moorfields Eye Hospital and University College London, and is not available for access beyond the AlzEye research team. National and international collaborations are welcomed, although restrictions on access to the cohort mean that only the AlzEye researchers can directly analyse individual-level systemic health data. More details can be found on the paper
PubMed id: 37704728
Altmetrics provided by Altmetric
