Automating the Process of Detection and Diagnosis of Glaucoma Using Deep Learning Approach

Abstract

Glaucoma is an eye disease. It is one of the reasons for impaired vision. Revitalization of the worsened tissues present in the optic nerve is difficult, so the detection of glaucoma in its early stage is very much essential. The proposed method first enhances the given image and identifies the presence of glaucoma. The method aims to identify the anterior chamber location for effectual angle calculation. If the angle is found to be greater than the threshold angle, then the eye is categorized as a normal eye else as glaucoma affected eye. The 250 ultrasound images collected from local eye hospitals were used as data for the experiment. The performance of the proposed method is considered in terms of classification accuracy and overall 95.9% result is achieved with 5 fold cross-validation setup respectively. This research work showed that using image processing techniques and deep learning achieved good performance for automatic glaucoma identification.