Investigative Ophthalmology & Visual Science Cover Image for Volume 65, Issue 7
June 2024
Volume 65, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2024
Evaluating RETFound for out-of-domain generalization in multi-disease detection from color fundus photographs
Author Affiliations & Notes
  • Sarah Matta
    Universite de Bretagne Occidentale, Brest, Bretagne, France
    INSERM, UMR 1101, Brest, F-29200, France
  • Mathieu Lamard
    Universite de Bretagne Occidentale, Brest, Bretagne, France
  • Alexandre Le Guilcher
    Evolucare Technologies, France
  • Laurent Borderie
    Evolucare Technologies, France
  • Pascale Massin
    Assistance Publique - Hopitaux de Paris, Paris, Île-de-France, France
  • Jean-Bernard Rottier
    Bâtiment de consultation porte 14 Pôle Santé Sud CMCM, Le Mans, F-72100, France
  • Béatrice Cochener
    Universite de Bretagne Occidentale, Brest, Bretagne, France
    INSERM, UMR 1101, Brest, F-29200, France
  • Gwenolé Quellec
    INSERM, UMR 1101, Brest, F-29200, France
  • Footnotes
    Commercial Relationships   Sarah Matta None; Mathieu Lamard None; Alexandre Le Guilcher Evolucare Technologies, Code E (Employment), OphtAI, Code E (Employment); Laurent Borderie Evolucare Technologies, Code E (Employment); Pascale Massin Allerga, Bayer, Novartis, Thea, Horus, Code C (Consultant/Contractor); Jean-Bernard Rottier None; Béatrice Cochener Thea, Alcon, Zeiss, B&L, Hoya, Horus, Santen, SIFI, Cutting Edge, J&J, Code C (Consultant/Contractor); Gwenolé Quellec Evolucare Technologies, Adcis, Code C (Consultant/Contractor)
  • Footnotes
    Support  This work received state aid managed by the National Research Agency under the LabCom program (ANR-19-LCV2-0005 - ADMIRE project).
Investigative Ophthalmology & Visual Science June 2024, Vol.65, 2336. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sarah Matta, Mathieu Lamard, Alexandre Le Guilcher, Laurent Borderie, Pascale Massin, Jean-Bernard Rottier, Béatrice Cochener, Gwenolé Quellec; Evaluating RETFound for out-of-domain generalization in multi-disease detection from color fundus photographs. Invest. Ophthalmol. Vis. Sci. 2024;65(7):2336.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : RETFound, a foundation model trained on 1.6 million unlabeled retinal images using Self-Supervised Learning (SSL), was recently introduced to boost models trained with minimal labeled data. It showed promising results: diabetic retinopathy classifiers for Color Fundus Photographs (CFP) better generalize to unseen datasets when pre-trained with RETFound. More generally, in this study, we assess the out-of-domain generalizability of multi-disease detection models for CFP, when pretrained with RETFound.

Methods : Four CFP datasets were considered: OPHDIAT (France, diabetic population, 77,827 images), OphtaMaine (France, general population, 17,120 images), RIADD (India, general population, 3,200 images) and ODIR (China, general population, 10,000 images). 7 disease categories were targeted: Diabetes, Glaucoma, Cataract, AMD, Hypertension, Myopia and Others. Cross-dataset evaluation was conducted: RETFound was fine-tuned for multi-disease detection on one dataset and evaluated on the others. RETFound was compared with two pre-trained models sharing the same architecture (ViT), but trained on ImageNet: one using Supervised Learning (SL-ImageNet), and the other using SSL (SSL-ImageNet). In addition, we compared SL-ImageNet with SL-bestArch-ImageNet, also pretrained through SL on ImageNet, but using the best possible architecture. A paired samples Wilcoxon test with Bonferroni correction was conducted to compare the per-class Area Under the receiver operating characteristic Curve (AUC) of each pretraining strategy.

Results : On out-of-domain datasets, the median per-category AUC value was 0.8207, 0.7296, 0.7646 and 0.8399, when fine-tuning RETFound, SL-ImageNet, SSL-ImageNet and SL-bestArch-ImageNet, respectively; the best architecture for SL was efficientnet-b5-ns. RETFound achieved significantly higher performances when compared to SSL-ImageNet (p=0.0044) and to SL-ImageNet (p=7.4e-07). However, SL-bestArch-ImageNet significantly outperformed SL-ImageNet (p=4.8e-06).

Conclusions : This study demonstrates the superiority of out-of-domain generalization performances of RETFound for multi-disease detection in CFP, in comparison to SL or SSL pretraining on ImageNet. It highlights that ViT is not the best architecture for this task, suggesting that improvement could be achieved by building a foundation model for different architectures.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Boxplot of AUCs across different models. ***p<0.001,**p<0.01

Boxplot of AUCs across different models. ***p<0.001,**p<0.01

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×