Predicting the risk of developing diabetic retinopathy using deep learning

Ashish Bora,Siva Balasubramanian,Boris Babenko,Sunny Virmani,Subhashini Venugopalan,Akinori Mitani,Guilherme De Oliveira Marinho,Jorge Cuadros,Paisan Ruamviboonsuk,Greg S Corrado,Lily Peng,Dale R Webster,Avinash V Varadarajan,Naama Hammel,Yun Liu,Pinal Bavishi

doi:10.1016/s2589-7500(20)30250-8

Ashish Bora, Siva Balasubramanian + Show 14 more

Open Access

https://doi.org/10.1016/s2589-7500(20)30250-8

Copy DOI

Abstract

Diabetic retinopathy screening is instrumental to preventing blindness, but scaling up screening is challenging because of the increasing number of patients with all forms of diabetes. We aimed to create a deep-learning system to predict the risk of patients with diabetes developing diabetic retinopathy within 2 years. We created and validated two versions of a deep-learning system to predict the development of diabetic retinopathy in patients with diabetes who had had teleretinal diabetic retinopathy screening in a primary care setting. The input for the two versions was either a set of three-field or one-field colour fundus photographs. Of the 575 431 eyes in the development set 28 899 had known outcomes, with the remaining 546 532 eyes used to augment the training process via multitask learning. Validation was done on one eye (selected at random) per patient from two datasets: an internal validation (from EyePACS, a teleretinal screening service in the USA) set of 3678 eyes with known outcomes and an external validation (from Thailand) set of 2345 eyes with known outcomes. The three-field deep-learning system had an area under the receiver operating characteristic curve (AUC) of 0·79 (95% CI 0·77-0·81) in the internal validation set. Assessment of the external validation set-which contained only one-field colour fundus photographs-with the one-field deep-learning system gave an AUC of 0·70 (0·67-0·74). In the internal validation set, the AUC of available risk factors was 0·72 (0·68-0·76), which improved to 0·81 (0·77-0·84) after combining the deep-learning system with these risk factors (p<0·0001). In the external validation set, the corresponding AUC improved from 0·62 (0·58-0·66) to 0·71 (0·68-0·75; p<0·0001) following the addition of the deep-learning system to available risk factors. The deep-learning systems predicted diabetic retinopathy development using colour fundus photographs, and the systems were independent of and more informative than available risk factors. Such a risk stratification tool might help to optimise screening intervals to reduce costs while improving vision-related outcomes. Google.

Highlights

Diabetic retinopathy is the leading cause of preventable blindness in adults aged 20–74 years.[1]
With the goal of early detection, regular screening is recommended by major organisations—including the American Diabetes Association,[2] International Council of Ophthalmology,[3] and American Academy of Ophthalmology4—at intervals ranging from every 12 months to 24 months for patients with no or mild diabetic retinopathy
The system was evaluated using an internal vali dation set of 7976 eyes and an external validation set of 4762 eyes

Summary

Introduction

Diabetic retinopathy is the leading cause of preventable blindness in adults aged 20–74 years.[1]. Regular screening is crucial to preventing blindness, the expected increase in the number of patients with diabetes—from 415 million in 2015 to a predicted 642 million in 20405— means that the burden of screening and follow-up represent a substantial challenge. The efficiency of diabetic retinopathy screening programmes might be improved by personalising screening frequencies on the basis of the likelihood of the development or progression of diabetic retinopathy.[6] We created a deep-learning system that uses colour fundus photographs to predict the risk of developing diabetic retinopathy. Signs of retinal microvascular changes caused by dia betes are visible in colour fundus photographs, which are routinely used to assess the stage of diabetic retino pathy. Modifiable risk factors include hyperglycaemia, hypertension, dyslipidaemia and obe sity, smoking, anaemia, pregnancy, low health literacy, inadequate access to health care, and poor adherence to therapy.[7,8] Non-modifiable risk factors include ethnicity, family history or genetics, age at onset of diabetes, type of diabetes, and duration of diabetes.[7]

Objectives

Methods

Results

Conclusion