Glaucoma Diagnosis Based on Fusion of Shape and Texture Features from Neuroretinal Rim Area of Retinal Fundus Image

Glaucoma diagnosis mainly involves parameters like optic cup-to-disc (CDR) measurement and intraocular pressure (IOP), whereas another important less discussed clinical biomarker is the neuroretinal rim (NRR). The categorization or classification of glaucoma through the machine learning applications significantly depends on the chosen feature extraction technique, feature selection process and the type of classifier utilized. This paper proposes a machine learning based glaucoma diagnosis method using the fusion of shape and texture features which are extracted from the neuroretinal rim areas of the optic nerve head fundus images. The data used in this study consists of 650 (168 glaucoma affected and 482 healthy) retinal fundus images. Random forest algorithm is used as a classifier to differentiate between the glaucoma and the healthy eyes. The performance obtained with the proposed approach is measured in terms of accuracy, sensitivity and specificity with an accuracy of 66.15%, a sensitivity of 18.45%, and a specificity of 82.78% respectively.