Detection of Nuclear Cataract in Retinal Fundus Image using RadialBasis FunctionbasedSVM

Abstract

Nuclear Cataract is a common eye disease that generally occurs at elder age. But if it's not detected at its earlier state, then it may affect vision and can live permanently. In this work, to detect the cataract an automated model proposed based on image processing and machine learning techniques. The input to the proposed model is, a set of fundus retinal images. For training the model, the image dataset consists of two types ofimages healthy and cataract affected. From each input retinal image a binary image, consisting of blood vessels is generated, using image processing techniques like image Filtration, segmentation and thresholding. These set of binary images are used as the feature matrix for defining the classifier by using a well-known machine learning technique Support vector machine (SVM). For validation and compression of the model, different kernels of SVM like linear, polynomial and RBF are applied and tested. Out of all, Radial Basis Function (RBF) based SVM performs good with an overall accuracy of 95.2 % and able to produce result in real time.