WITHDRAWN: Segmentation of retinal vasculature through composite features and supervised learning

Abstract

Analysis of Retinal fundus images has obtained a significant interest in research due to its widespread applicability in the detection of various diseases related to the eye. This paper focused in the analysis of Diabetic Retinopathy through different features (Optic Disk, Retinal Vessels, and Exudates etc.,) of retinal image. Towards this objective, this paper presents a new process for segmenting retinal vessels. The proposed mechanism accomplished the Gabor Filter for Extracting Features and Support Vector Machine Algorithm for classification. Here the Gabor Filter ensures a more resilience to the scaling and orientation issues in the retinal image. Afterwards, a feature set consists of thirteen features is extracted from retinal image to provide a proper differentiation between the image pixels and background pixels. Based on these features, the SVM classifier classifies the vessel pixels and background pixels more effectively which improves the classification accuracy and reduces false positive rate. An extensive simulation is carried out over the developed approach through two benchmark and freely accessible retinal image datasets; they are DRIVE and STARE. The obtained results reveal the outstanding performance with respect to the performance metrics such as accuracy, sensitivity and specificity.