A lightweight CNN model for early detection of Glaucoma

Glaucoma is rapidly increasing in middle-aged population in India and worldwide. High intra-ocular pressure damages the optic nerve and progresses towards loss of vision an irreversible process. Hence, early detection of the disease can save people from losing their faculty of vision. Various techniques in the literature, such as threshold-based segmentation of optic disc (OD) and optic cup (OC), super pixel approaches, deep Convolutional generative adversarial networks, optical coherence tomography (OCT), and high- resolution tomography (HRT), are either computationally complex or time-consuming. This study proposes a lightweight Convolutional Neural Network (CNN) implemented in MATLAB, achieving the highest accuracy on popular glaucoma datasets. A peak accuracy of 99.8% has been achieved with the proposed lightweight CNN model