Glaucoma Disease Classification based on Convolutional Neural Network Architecture Method

Abstract

Glaucoma refers to a class of eye disorders which often associated with the increase in intraocular pressure that can damage the eye by destroying the nerve cells in the retina in addition to the optic nerve of the eye. The need for detecting and recognizing glaucoma eye disease increased worldwide. The current paper presents a novel Convolutional Neural Network (CNN) based on an intelligent pattern categorization method upon ACRIMA dataset images. The analysis involved using 95% of the dataset in training and 5% for testing. The proposed technique for detecting and classifying the glaucoma in fundus images consisted of a multi-layer deep neural network model with convolutional and classification layers. The originality of the current study aims to verify the impact of the number of filters (nf) upon 50 and 100 Epoch values in classifying glaucoma disease. Results gave a state-of-art performance in classifying glaucoma and recorded a unity value for sensitivity, 99.35% for specificity, and 99.29% for accuracy. The variation of the number of filters recorded higher values for all estimators, and hence the proposed model at final succeeded with significant advancement in comparison with other related studies.