Modified 2D-EWT-based Automated System for Glaucoma Diagnosis

Glaucoma is an eye disease which has been one of the leading causes of loss of vision worldwide. It occurs when the fluid pressure in the optic nerve increases which causes the damage to the optic nerve. Detection of glaucoma is necessary for proper and timely treatment. In this paper, an automated system is developed for glaucoma diagnosis in which a two-dimensional empirical wavelet transform (2D-EWT)-based fundus image decomposition method is developed. A novel method is suggested for magnitude spectrum-based image segmentation. In 2D-EWT, the average magnitude spectrum along the horizontal or vertical axis is considered which discards the information along one axis. In the suggested method, the average magnitude spectrum along both axes is utilized and more decomposed images can be obtained. The decomposed fundus images are given to the ResNet-18-based deep learning model for glaucoma identification. The performance of the proposed method is computed in terms of accuracy (ACC), sensitivity (SEN), and specificity (SPEC). The proposed method has achieved the ACC of 99.33% with 99.35% SEN, and 99.25% SPEC which is more than the compared methods.