Glaucoma Identification on Fundus Retinal Images Using Statistical Modelling Approach

Abstract

Abstract Glaucoma is an eye disease characterized by progressive deterioration of the optic nerve head and a broad view that can cause blindness. The Population Based Survey in 2010 indicates that glaucoma was the second leading cause of blindness after cataracts, which was about 8% of 36 million sufferers of blindness worldwide. Symptoms of glaucoma that arise usually cannot be felt directly. So it is necessary to do an eye examination to find out glaucoma, one of which is to look at the size of the optic disk in the digital fundus photo. The previous studies about glaucoma identification were done by using mathematical computation approach that have still not satisfied. Therefore, in this study we propose a new method, i.e., statistical modelling approach to identify glaucoma. In statistical modelling, there are two approaches, i.e., parametrical approach, and non-parametrical approach based on penalized spline estimator. The result of classification accuracy by using parametrical and non-parametrical approaches are 73.3% and 93.33%, respectively. Based on the result, we conclude that non-parametrical approach has better outcome so that it can be used to identify glaucoma on fundus retinal image.