Description

Diabetic retinopathy is a most commonly occurring disease nowadays. Early identification of the disease will help in the preserving the eye sight in many patients. For the identification of the diseases the classifiers were used. Features were extracted from the images that were given as the input to the classifiers. Features were extracted from the images based on numerous methodologies. The main aim of extracting features is to reduce the dimensions of the original data. The shape based features were more efficient in the identification disease because in the fundus images the diseases that were occurring due to the diabetic retinopathy present in fundus images has specific intensity differences and shapes. The classification is employed to identify whether the input image has diseases or not. Three different type of classifiers were chosen for comparison. The performance of the classifiers were measured based on the performance metrics and the comparisons were done. ROC analysis and confusion matrix were selected for the performance metrics. The ROC analysis is a plot of the sensitivity and the specificity of the classifiers. The confusion matrix is an overall representation of the performance of the classifier.