A foundation model for generalizable disease detection from retinal images.

Journal: Nature
Published Date: Sep 13, 2023

Abstract

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. However, the development of AI models requires substantial annotation and models are usually task-specific with limited generalizability to different clinical applications. 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.

Authors

  • Yukun Zhou
    Centre for Medical Image Computing, University College London, London, UK.
  • Mark A Chia
    NIHR Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, UK.
  • Siegfried K Wagner
    National Institute for Health Research Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, UK.
  • Murat S Ayhan
    Centre for Medical Image Computing, University College London, London, UK.
  • Dominic J Williamson
    Centre for Medical Image Computing, University College London, London, UK.
  • Robbert R Struyven
    Centre for Medical Image Computing, University College London, London, UK.
  • Timing Liu
    NIHR Biomedical Research Centre Fellow, University College London, London, UK.
  • Moucheng Xu
    Centre for Medical Image Computing, University College London, London, UK.
  • Mateo G Lozano
    NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust, London, UK.
  • Peter Woodward-Court
    NIHR Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, UK.
  • Yuka Kihara
    Department of Ophthalmology, University of Washington, Seattle.
  • Andre Altmann
    Centre for Medical Image Computing, University College London, London, UK.
  • Aaron Y Lee
    Department of Ophthalmology, University of Washington, Seattle, Washington.
  • Eric J Topol
    Scripps Research Translational Institute, La Jolla, CA 92037, USA; Scripps Clinic Division of Cardiovascular Diseases, La Jolla, CA 92037, USA. Electronic address: etopol@scripps.edu.
  • Alastair K Denniston
    Centre for Patient Reported Outcomes Research Institute of Applied Health Research University of Birmingham Birmingham Reino Unido Centre for Patient Reported Outcomes Research, Institute of Applied Health Research, University of Birmingham, Birmingham, Reino Unido.
  • Daniel C Alexander
    Centre for Medical Image Computing and Dept of Computer Science, University College London, Gower Street, London WC1E 6BT, UK.
  • Pearse A Keane
    National Institute for Health Research Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, UK.

Keywords

External Resources

Popular Topics
Recent Journals