Automated Diagnosis System of Diabetic Retinopathy Using GLCM Method and SVM Classifier

Abstract

Diabetic Retinopathy (DR) is the cause of blindness. Early identification needed for prevent the DR. However, High hospital cost for eye examination makes many patients allow the DR to spread and lead to blindness. This study identifies DR patients by using color fundus image with SVM classification method. The purpose of this study is to minimize the funds spent or can also be a breakthrough for people with DR who lack the funds for diagnosis in the hospital. Pre-processing process have a several steps such as green channel extraction, histogram equalization, filtering, optic disk removal with structuring elements on morphological operation, and contrast enhancement. Feature extraction of preprocessing result using GLCM and the data taken consists of contrast, correlation, energy, and homogeneity. The detected components in this study are blood vessels, microaneurysms, and hemorrhages. This study results what the accuracy of classification using SVM and feature from GLCM method is 82.35% for normal eye and DR, 100% for NPDR and PDR. So, this program can be used for diagnosing DR accurately.