Using deep learning to detect diabetic retinopathy on handheld non-mydriatic retinal images acquired by field workers in community settings

Abstract

Diabetic retinopathy (DR) at risk of vision loss (referable DR) needs to be identified by retinal screening and referred to an ophthalmologist. Existing automated algorithms have mostly been developed from images acquired with high cost mydriatic retinal cameras and cannot be applied in the settings used in most low- and middle-income countries. In this prospective multicentre study, we developed a deep learning system (DLS) that detects referable DR from retinal images acquired using handheld non-mydriatic fundus camera by non-technical field workers in 20 sites across India. Macula-centred and optic-disc-centred images from 16,247 eyes (9778 participants) were used to train and cross-validate the DLS and risk factor based logistic regression models. The DLS achieved an AUROC of 0.99 (1000 times bootstrapped 95% CI 0.98–0.99) using two-field retinal images, with 93.86 (91.34–96.08) sensitivity and 96.00 (94.68–98.09) specificity at the Youden’s index operational point. With single field inputs, the DLS reached AUROC of 0.98 (0.98–0.98) for the macula field and 0.96 (0.95–0.98) for the optic-disc field. Intergrader performance was 90.01 (88.95–91.01) sensitivity and 96.09 (95.72–96.42) specificity. The image based DLS outperformed all risk factor-based models. This DLS demonstrated a clinically acceptable performance for the identification of referable DR despite challenging image capture conditions.

Novel artificial intelligence algorithms for diabetic retinopathy and diabetic macular edema

“…In contrast, the SMART India Study Group described a DL model in detecting RDR using 2-field fundus photos acquired by nonmydriatic hand-held cameras from 16,247 eyes. The system achieved a high performance in detecting RDR, with an AUC of 0.98 and 0.99 with one-field and two-field inputs, respectively [ 67 ]. Significantly, they found that variations in dilation states and image gradability across studies could influence the results [ 65 , 67 , 68 ].…”

“…The system achieved a high performance in detecting RDR, with an AUC of 0.98 and 0.99 with one-field and two-field inputs, respectively [ 67 ]. Significantly, they found that variations in dilation states and image gradability across studies could influence the results [ 65 , 67 , 68 ]. In a recent study, the SELENA + algorithm (EyRIS Pte Ltd, Singapore), which was developed using traditional fundus photos, was integrated into a hand-held fundus camera and paralleled the results of conventional retina specialist evaluations, reinforcing its precision in DR detection in a different use setting [ 69 ].…”

Machine learning in precision diabetes care and cardiovascular risk prediction

“…Given their simplicity and wide availability, fundoscopic images were used in some of the earliest DL models in medicine, predicting diabetic retinopathy with performance matching that of expert readers [ 11 , 63 , 64 ]. This has opened the way for efficient screening of both diabetes, and diabetes-related chronic kidney disease and retinopathy in settings with limited resources, as demonstrated in real-world implementation studies in Thailand and India [ 65 , 66 ].…”

Evaluation of a portable retinal imaging device: towards a comparative quantitative analysis for morphological measurements of retinal blood vessels

“…A growing number of portable retinal imaging devices are on the market, frequently (but not exclusively) in the form of an add-on to a smartphone. Evaluation of several such devices has been reported in the setting of qualitative assessment of diabetic retinopathy and glaucoma [ 7 , 8 ], increasingly coupled with artificial intelligence approaches to classification [ 9 ]. However, the imaging requirements for the quantitative assessment of the retinal microvasculature are more demanding than for qualitative assessment of retinopathy, and it remains uncertain whether portable retinal imaging devices are suitable for this purpose.…”