Using Artificial Intelligence for Screening and Grading Diabetic Retinopathy Disease: An Overview

Diabetic retinopathy (DR) is a serious complication associated with diabetes that damages the delicate retinal membrane in the eye. The disease often develops silently, with no noticeable symptoms or mild vision impairment, and ultimately leads to blindness. Consequently, numerous studies have been conducted to detect and predict the onset of this disease at an earlier stage. These studies have employed various techniques to extract the indicators of DR and after, classify the disease into its four grades, utilizing color fundus images available in the datasets. This paper aims to provide essential definitions related to our research domain and comprehensively review the most relevant works employing different techniques in the sub-domains of artificial intelligence for feature extraction, classification, and prediction of DR. Moreover, it conducts a thorough comparative analysis of DR prediction methods, evaluating key performance measures. Additionally, this paper provides a detailed overview of the diverse datasets employed in related studies, and finally describes the architecture of our proposed system based on the CNN model.