Deep Learning-Based Ocular Disease Classification in Fundus Images

Ocular anomalies, such as glaucoma, cataracts, and age-related macular degeneration (AMD), represent a significant public health burden. In this paper, we propose a deep learning-based multi-class classification model for the accurate detection of these eye conditions using fundus images. The model utilizes a pre-trained ResNet-50 architecture and is trained and tested on a dataset of 1096 fundus images. Our results demonstrate the model’s effectiveness, achieving a classification accuracy of 97%. The precision, recall, and F1-score metrics show the model’s ability to make accurate predictions for each class, highlighting its robustness. Moreover, the model exhibits strong generalization capabilities and avoids overfitting, as evidenced by the training and validation accuracy and loss curves. The proposed model shows promise as a valuable tool for assisting medical professionals in the early and accurate diagnosis of ocular diseases, contributing to improved patient outcomes and ocular health management.