Detection of Diabetics in Retinal Images using Machine Learning Algorithms

Abstract

Normally nearly 90% of the old age person can suffer from Diabetics mellitus. The major cause of the diabetic’s mellitus is by the consuming of high sugar content food. So the people who havediabetic’s mellitus can suffer from diabetic retinopathy. The DR is the retina disease it can affect the retina of the human eye can cause mild vision or fully blindness if they not treated properly. Hemorrhages, hard Exudates and Micro aneurysms they are commonly called as HEM which can appear in the retina. So the HEM diagnosis has been taken to avoid the blindness. In the past the detection of the DR can be made by the texture features of the retina such as LBP. In this paper we propose the different set of texture features to predict the diabetic retinography in the earlier stage. The local Ternary Pattern and Local Energy-based Shape Histogram (LESH) are the two different set of features. The performance of the LBP can be projected in the method. The support vector machine (SVM) can be used in the classification and extraction of the texture. The texture features are get binned together. we also used other algorithms like random forest, gaussian, k neighbor. From the above methods random forest method is more accurate than other algorithms.