A Novel Method for Glaucoma Detection Using Computer Vision

Abstract

Glaucoma is an eye disorder that impairs the optic nerve and can cause permanent blindness if left untreated. This condition arises due to elevated pressure inside the retina. As damage from glaucoma is impossible to rectify, early detection helps to prevent vision loss. Manual examination of fundus photographs is a difficult process for medical examiners because a large set of images will have only a small number of glaucomatous images, so Computer-Aided systems can reduce the work load by automatically analyzing the images. Here a computer-Aided diagnosis system for glaucoma screening based on VCDR value along with texture features extracted using Gabor filter is proposed. The processed retinal fundus images are segmented and the VCDR value is calculated from the segmented images. The simulation is carried out in Python 3 and the databases used are DRISHTI GS1 and HRF. The system is evaluated using four classifiers SVM, KNN, Logistic regression and Random forest.