Red Lesion Detection Using Hough Transform and KNN Classifier for Diabetic Retinopathy Screening

Abstract

Diabetic retinopathy is a major cause of blindness in the world. It will take more time to identify the clinical features such as microaneurysms, hemorrhages, exudates and cottonwool spots through manual inspection of fundus images. A computer assisted diagnosis system can help to reduce the burden on the ophthalmologist and rapidly identify the most severe cases. An efficient approach for red lesion detection in fundus image is proposed. The fundus image is preprocessed and Hough transform is used to identify the regions having red lesions. A KNN classifier is used to train the feature which classifies the images according to their probability of being normal or with disease.

Severity Classification of Non-Proliferative Diabetic Retinopathy Using Support Vector Machine (SVM)

“…Several previous works have reported the automatic detection of DR using several machine-learning techniques. Some of these methods are the Support Vector Machine (SVM) method [7][8][9][10], the K-Nearest Neighbor (KNN) method [7,11,12], the Random Forest (RF) method [7,[13][14][15][16], Decision Tree (DT) method [7,17], Artificial Neural Network (ANN) method [18,19], and Probabilistic Neural Network (PNN) method [9].…”