Image Segmentation for Diabetic Retinopathy Using Modified Bacterial Foraging Optimization Algorithm Rajini N Hema* Assistant Professor, Department of Computer Science and Engineering, Alagappa Chettiar Government College of Engineering and Technology, Karaikudi, Tamilnadu, India *Corresponding Author: Hema Rajini N, Assistant Professor Department of Computer Science and Engineering, Alagappa Chettiar Government College of Engineering and Technology, Karaikudi-630003, Tamilnadu, India, Email: auhemasmith@yahoo.co.in
Online published on 19 August, 2019. Abstract In recent days, investigation of retinal vessels of fundus images is an essential way to screen and diagnose related diseases. Diabetes is a commonly occurring disease and is exponentially rising globally. diabetic retinopathy (DR)is an important reason for blindness. This paper presents a new image segmentation technique for DR. To segment the DR images, a modified bacterial foraging algorithm with Levi distribution. The efficiency of the (BFO-L) method is tested using two benchmark dataset namely DRIVE and STARE. A detailed comparative analysis is also made with the recently proposed methods under various measures. The experimental values imply that the BFO-L method detects tiny blood vessels and locates the edges effectively. While comparing the other methods on the applied DRIVE dataset, the presented method obtains maximum performance with a sensitivity value of 0.8725, specificity value of 0.9314, accuracy of 0.9798 and AUC of 0.9895 respectively. Top Keywords DR, Image Segmentation, Levi distribution, Retina. Top |