An Automatic Detection of Blood Vessel in Retinal Images Using Convolution Neural Network for Diabetic Retinopathy Detection

Abstract

Diabetes is a typical chronic disease that may remind to numerous complications. Since the diabetic patients, the diabetic retinopathy (DR) is standout amongst the most serious of these inconveniences and also most steady reasons of vision loss. Automatic detection of diabetic retinopathy at early stage is helping the ophthalmologist to treat the affected patient and avoid vision loss. Therefore, in this paper, we develop an efficient automatic diabetic detection in retinal images using convolution neural network. The suggested system mainly comprises of five modules such as (i) preprocessing, (ii) blood vessel segmentation, (iii) exudates segmentation, (iv) texture feature extraction, and (v) diabetic detection. At first, the preprocessing step is carried out using adaptive histogram equalization (AHE) for enhancing the input retinal image. Consequently, blood vessel segmentation and exudates segmentation are done using convolution neural network (CNN) and fuzzy c-means clustering (FCM) respectively. Then, texture features are extracted from blood vessel and exudates. After the feature extraction, the diabetic classification is done with the help of support vector machine. The experimental results demonstrate that the proposed approach accomplishes better diabetic detection result (accuracy, sensitivity, and specificity) compared to other approaches.