A comparative evaluation of diabetic retinal vascular structures using edge detection techniques

Abstract

Diabetic retinopathy (DR) is a major cause of loss of vision in diabetes which occurs due to vascular changes in the retina. Automated image processing has the potential to assist in the early detection of diabetes, by detecting changes in blood vessel patterns in the retina. Accurate extraction of retinal blood vessels is an important task in computer aided diagnosis of DR. The edge detection technique has been greatly benefited in interpreting the information contents in the retinal blood vessel. The pre-processing of retinal image may help in detecting the early stage of symptoms in DR. In this article we have detected the abnormalities in retinal blood vessels by Sobel and Canny edge detector and the results are compared in terms of different image parameters and histogram error. The Canny operator is found to have higher peak signal-to-noise ratio (PSNR) and normalised cross-correlation (NK) with less value of mean square error (MSE), average difference (AD), and structural content (SC). The histogram error is also found to be less in the Canny operator (0.0015) than the Sobel operator (0.014). Thus, we may conclude that the Canny operator is more accurate in detecting even tiny blood vessels compared to Sobel operator in diabetic retinal image.