Retinal blood vessel segmentation using edge detection method

Abstract

Currently eye disease is a major challenge in our life. The main diseases in human eye are Ocular Hypertension, Glaucoma, Diabetic Muscular Edema, Diabetic Retinopathy, Color Blindness, Phonetic, Blindness etc. The only way is to identify the cause and then the treatment could be stared as early as possible. To detect in 1st stage, the measurement as well as automatic tracking of the blood vessels of human eye are very essential task. The photograph of human eye is called Fundus image. To detect the eye vessels, segmentation of these vessels are the main motivation of the work. Here, a survey is being done to do the detection using various edge based segmentation procedures, i.e., Kirsch filter, Canny, Prewitt, Sobel and Fuzzy-C. The work is applied on freely available DRIVE database. The result shows that, Kirsch filter is the best among the other benchmark methods when the results are evaluated for the parameters such as, Accuracy, Specificity and Sensitivity. The outcome of Accuracy is varies from 0.77 to 0.94, Specificity varies in the range between 0.76 to 1.00 and finally Sensitivity lies between 0.20 to 0.84.