Eminent identification and classification of Diabetic Retinopathy in clinical fundus images using Probabilistic Neural Network

Abstract

Diabetic retinopathy occurs due to increased risk at the eye via swelling and occurrence of abnormalities in blood vessels. In earlier several methods have been implemented to detect the diabetic retinopathy, but there is a lack of efficiency over the traditional methods that utilize unsupervised algorithms. Recently, deep learning plays a significant role in several technological applications such as Image recognition and semantic analysis that utilizes for portraying the DR. In our proposed methodology, an innovative framework has been implemented to overcome the issues of traditional methodology. Initially, an input image is acquired from the dataset. Then the preprocessing is performed where the image resizing and the double conversion is carried out and next to this image segmentation is the process where the morphological and thresholding-based operations are performed. The GLCM - an effective feature is chosen for extracting the features with the co-occurrence matrix. After extracting the features, the classification process is performed using Probabilistic Neural Network (PNN) which provides an effective classifier output. It is concluded that this novel vessel segmentation framework acquired better accuracy, sensitivity, F measures, specificity and precision from this experiment.