A Deep Learning Framework for Analyzing Retinal Fundus Images and Classifying Diseases

Cataracts are one of the leading causes of blindness and vision impairment worldwide, particularly in the elderly. Early identification of cataract severity is crucial for planning appropriate medical intervention and preventing further visual deterioration. Traditional cataract diagnosis relies heavily on the manual examinations performed by ophthalmologists, which can be time-consuming and subjective. This paper proposes an automated deep learning technique that classifies cataract phases into immature and mature categories using ocular images to address this problem. A Convolutional Neural Network (CNN) model is developed to learn discriminative visual characteristics from eye images following appropriate preprocessing and normalisation. The annotated images in the dataset that show both stages of cataracts allow the model to be trained under supervision. The trained network is evaluated using performance metrics such as accuracy, precision, recall, and F1-score. The studies' findings demonstrate that the proposed approach effectively and consistently distinguishes between immature and mature cataracts. Furthermore, the model is integrated into a user-friendly Flask-based online application that allows real-time prediction and image submission. The proposed framework provides an accessible and cost-effective means to assist ophthalmologists and encourage early cataract screening in distant and resource-constrained settings. Keywords :Cataract Detection, Deep Learning, Convolutional Neural Network (CNN), Medical Image Classification, Ocular Image Analysis, Healthcare Artificial Intelligence, Automated Disease Diagnosis