Deep Learning Model for Diabetic Retinopathy Detection Using MobileNet Architecture

Diabetic Retinopathy (DR) is a significant complication of diabetes mellitus. Its prevalence in Sri Lanka is increasing due to limited public awareness and delays in diagnosis caused by resource constraints in the healthcare system. To facilitate early and efficient detection, this study proposes a deep learning model for DR detection in Sri Lankan patients. A dataset of 272 retinal fundus images was collected from a Sri Lankan hospital and used to train a model based on the MobileNet architecture, achieving an accuracy of 99.19%. The study faced limitations due to the small dataset size. Future work will focus on collecting retinal fundus images from across the island to enhance the dataset and improve model performance.