Dr. ASTHA JAIN

Dr. S.NATARAJAN, Dr.Radhika Krishnan

Abstract

The aim of the study was to validate the use of offline artificial intelligence for community screening of diabetic retinopathy. Diabetics above 40 years age, visiting dispensaries run by municipal corporation of Mumbai were included in the study. Three field fundus photo was done using smartphone based fundus camera with integrated offline Artificial intelligence (AI) algorithm. The images were also analysed by a vitreoretinal surgeon. A total of 1378 patients consented for the study. Gradable fundus images were obtained and analysed for 1294 patients. The sensitivity and specificity of AI for diagnosing referrable diabetic retinopathy was 100% and 89.55% respectively; the same values for any DR were 89·13% and 94·43% respectively with no false negative results. The robustness of the offline AI algorithm was established in this study making it a reliable tool for community-based DR screening.

Full Text