Ocular Disease Classification Using CNN with Deep Convolutional Generative Adversarial Network

The Convolutional Neural Network (CNN) has shown promising performance in image classification because of its robust learning potentialities. However, it demands a substantial and balanced dataset for effective training. Otherwise, networks frequently exhibit overfitting and struggle to generalize to new examples. A publicly available dataset of fundus images of ocular disease is insufficient to train any classification model to achieve satisfactory accuracy. So, we propose a Generative adversarial Network (GAN) based data generation technique to synthesize a dataset for training a CNN-based classification model and later use the original disease containing ocular images to test the classification accuracy. During testing the model classification accuracy with the original ocular image, the proposed method attained an accuracy rate of 78.6% for myopia, 88.6% for glaucoma, and 84.6% for cataracts, with an overall average classification accuracy of 84.6%.