Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes

Abstract

Importance: A deep learning system (DLS) is a machine learning technology with potential for screening diabetic retinopathy and related eye diseases.

Objective: To evaluate the performance of a DLS in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, possible glaucoma, and age-related macular degeneration (AMD) in community and clinic-based multiethnic populations with diabetes.

Design, setting, and participants: Diagnostic performance of a DLS for diabetic retinopathy and related eye diseases was evaluated using 494 661 retinal images. A DLS was trained for detecting diabetic retinopathy (using 76 370 images), possible glaucoma (125 189 images), and AMD (72 610 images), and performance of DLS was evaluated for detecting diabetic retinopathy (using 112 648 images), possible glaucoma (71 896 images), and AMD (35 948 images). Training of the DLS was completed in May 2016, and validation of the DLS was completed in May 2017 for detection of referable diabetic retinopathy (moderate nonproliferative diabetic retinopathy or worse) and vision-threatening diabetic retinopathy (severe nonproliferative diabetic retinopathy or worse) using a primary validation data set in the Singapore National Diabetic Retinopathy Screening Program and 10 multiethnic cohorts with diabetes.

Exposures: Use of a deep learning system.

Main outcomes and measures: Area under the receiver operating characteristic curve (AUC) and sensitivity and specificity of the DLS with professional graders (retinal specialists, general ophthalmologists, trained graders, or optometrists) as the reference standard.

Results: In the primary validation dataset (n = 14 880 patients; 71 896 images; mean [SD] age, 60.2 [2.2] years; 54.6% men), the prevalence of referable diabetic retinopathy was 3.0%; vision-threatening diabetic retinopathy, 0.6%; possible glaucoma, 0.1%; and AMD, 2.5%. The AUC of the DLS for referable diabetic retinopathy was 0.936 (95% CI, 0.925-0.943), sensitivity was 90.5% (95% CI, 87.3%-93.0%), and specificity was 91.6% (95% CI, 91.0%-92.2%). For vision-threatening diabetic retinopathy, AUC was 0.958 (95% CI, 0.956-0.961), sensitivity was 100% (95% CI, 94.1%-100.0%), and specificity was 91.1% (95% CI, 90.7%-91.4%). For possible glaucoma, AUC was 0.942 (95% CI, 0.929-0.954), sensitivity was 96.4% (95% CI, 81.7%-99.9%), and specificity was 87.2% (95% CI, 86.8%-87.5%). For AMD, AUC was 0.931 (95% CI, 0.928-0.935), sensitivity was 93.2% (95% CI, 91.1%-99.8%), and specificity was 88.7% (95% CI, 88.3%-89.0%). For referable diabetic retinopathy in the 10 additional datasets, AUC range was 0.889 to 0.983 (n = 40 752 images).

Conclusions and relevance: In this evaluation of retinal images from multiethnic cohorts of patients with diabetes, the DLS had high sensitivity and specificity for identifying diabetic retinopathy and related eye diseases. Further research is necessary to evaluate the applicability of the DLS in health care settings and the utility of the DLS to improve vision outcomes.

Figure 1.. Receiver Operating Characteristic Curve and Area Under the Curve of the Deep Learning System for Detection of Referable Diabetic Retinopathy and Vision-Threatening Diabetic Retinopathy in the Singapore National Diabetic Retinopathy Screening Program (SIDRP 2014-2015; Primary Validation Dataset), Compared with Professional Graders’ Performance, With Retinal Specialists’ Grading as Reference Standard

AUC indicates area under the receiver operating characteristic curve; SIDRP, Singapore National Diabetic Retinopathy Screening Program.

Figure 2.. Receiver Operating Characteristic Curve and Area Under the Curve of the Deep Learning System for Detection of Referable Diabetic Retinopathy in SIDRP 2014-2015 (Primary Validation Set) by Age, Sex, and HbA_1c Level

Eyes are the units of analysis. Glycated hemoglobin (HbA_1c) levels were available for only 52.1% of patients. Cluster-bootstrap biased-corrected 95% CI was computed for each area under the receiver operating characteristic curve (AUC), with individual patients as the bootstrap sampling clusters. See Methods for defintions of referable conditions. A, P < .001. B, P = .74. C, P = .34. SIDRP indicates Singapore National Diabetic Retinopathy Screening Program.

Figure 3.. Primary Validation Dataset and Area Under the Curve of the Deep Learning System in Detecting Referable Possible Glaucoma and Referable Age-Related Macular Degeneration (AMD) Among Patients With Diabetes, SIDRP 2014-2015, With Reference to a Retinal Specialist

Eyes are the units of analysis. Cluster-bootstrap biased-corrected 95% CI was computed for each area under the receiver operating characteristic curve (AUC), with individual patients as the bootstrap sampling clusters. Referable possible glaucoma defined as ratio of vertical cup to disc diameter of 0.8 or greater, focal thinning or notching of the neuroretinal rim, optic disc hemorrhages, or localized retinal nerve fiber layer defects. Referable acute macular degeneration (AMD) defined as the presence of intermediate AMD (numerous intermediate drusens, 1 large drusen >125um) and/or advanced AMD, geographic atrophy, or neovascular AMD, using the Age-Related Eye Disease Study grading system. Repeats from the Singapore National Diabetes Retinopathy Screening Program (SIDRP) 2014-2015 were excluded from the analysis. Asymptotic 95% CI was computed for the logit of each proportion and using the cluster sandwich estimator of standard error to account for possible dependency of eyes within each individual. Cluster-bootstrap biased-corrected 95% CI was computed for each AUC, with individual patients as the bootstrap sampling clusters.