Automated Diagnosis of Glaucoma by Using 1-D Retina Images

Abstract

Glaucoma is an eye disease associated that a worldwide leading cause of irreversible vision loss. This disease is caused by an increase in the pressure in the aqueous humor. Glaucoma can be asymptomatic until a relatively late stage and for this reason, diagnosis is frequently delayed. Early detection of the disease is the key to preventing the progression of the disease and reducing the patient’s vision. This paper investigates an approach for Glaucoma detection using retinal images. In this paper, after the pre-processing level and vessel segmentation of retinal images, the features are extracted by the fractional Fourier transform. Then, the feature’s matrix of 2-D images converted to 1-D images. This paper used the receiver operating characteristic curve for feature selection. Finally, the selected features are classified by different classifiers. We used the high-resolution fundus Glaucoma retina images databases. We achieved an accuracy of 100% with a support vector machine classifier and a linear discriminant analysis classifier. The results indicate using the proposed algorithm, Glaucoma can be detected fast and accurately.