Analysis of Retinal Fundus Images for Classification of Glaucoma

An early sign of glaucoma can be seen from the presence of damage to the retinal nerve fibers. In retinal fundus image, the damage of nerve fibers is represented as a dark area. Several studies have been done with various analyses, one of those used texture. Thus, this paper aims to classify glaucomatous and healthy images by analyzing textures using first-order statistical approach. Initially, the image is cropped to get the region of interest (ROI), then the blood vessels and optic disc are segmented and removed to get candidate image. The features of candidate image will be extracted using first-order statistic. To obtain high accuracy, all features are evaluated by using Relief to select the efficient features. Selected features are used to classify the data into two classes. In this work, support vector machine (SVM), k-nearest neighbor (k-NN), linear discriminant analysis (LDA), and Naive Bayes are used. The performance results show that k-NN performs better with average of accuracy, sensitivity, specificity, PPV, and NPV are 93.3% than the others that are less than 90%.