ABSTRACT
Diabetic Retinopathy (DR) is identified as one of the leading causes loss of vision for the disease's sufferers. Thus, proper management via early detection is greatly needed. In this study, a deep learning–based approach is proposed for the automated detection of diabetic retinopathy using the InceptionV3 model. The dataset used to train the model was APTOS 2019 Blindness Detection. This is a collection of images of retinal fundus, labeled across five stages of DR. It is the capability of InceptionV3 for extracting robust features, and what the model attempted to use in an effort to learn the complex patterns in images about the extent of the disease. This model has an accuracy of 74%, which suggests potential toward early detection and entirely automated screening of the retina. Such an approach may offload some of the workload on healthcare professionals, provide quicker diagnoses, and allow an improvement in efficiency levels regarding diabetic retinopathy detection, especially in resource-poor setups.
