Diabetic Retinal Disease Detection Through Transfer Learning Techniques

Visual impairment is becoming increasingly common across the globe, necessitating advanced diagnostic methods. Conventional diagnostic methods are largely dependent on manual expertise. However, recent advancements in Machine Learning (ML) and Artificial Intelligence (AI) research, especially in medical diagnosis, have shown promise. Ophthalmologists rely on fundus and Optical Coherence Tomography (OCT) images as essential tools for diagnosing retinal disorders. Detecting eye disorders early greatly enhances the likelihood of recovery and helps prevent visual impairment. To diagnose and prevent various eye problems, it is beneficial to have an intelligent computerized method for classifying them. Previous studies have classified patients as either healthy or affected by eye diseases. This research paper presents an expert system based on ML and AI that not only classifies eye diseases, but also determines the stage or type of disease and provides the percentage probability of classification. The system features a simple and fast network architecture capable of delivering precise disease stage/type identification and probability percentages, enhancing its utility in clinical settings. All three models achieved accuracies above 81% post training.