Glaucoma Retinal Image Detection and Classification using Machine Learning Algorithms

Abstract

The diseases correlated with retina are categorized into Diabetic Retinopathy (DR) and Glaucoma. Glaucoma is irreversible and one of the leading causes of blindness and it is very important to detect in its early stage because late diagnose will result in permanent vision loss. It is mainly characterized by the malfunctioning of ganglion cells, which changes the structure of the optic nerve head and thickness of retinal nerve fiber layer. Therefore, it is very important to detect the Glaucoma in order to prevent earlier vision loss. In this paper, the Glaucoma disease is detected using various machine learning classification algorithms Support Vector Machine (SVM), Neural Network (NN) and Adaptive Neuro Fuzzy Inference System (ANFIS) classifiers. These classifiers are used to classify the source retinal image into either normal or abnormal. The proposed methods are applied and tested on the retinal images which are available from Retinal fundus images for Glaucoma Analysis (RIGA) and High-Resolution Fundus (HRF) dataset. The Glaucoma detection method using ANFIS classifier obtains 97.2% of Precision, 97.3% of Recall and 97.1% Accuracy.