Publish In
International Journal of Advance Computational Engineering and Networking (IJACEN)-IJACEN
Journal Home
Volume Issue
Volume-2,Issue-3  ( Mar, 2014 )
Paper Title
Pre Processing Of Vessel Segmentation For The Identification Of Cardiovascular Diseases With Retinal Images
Author Name
V. Jefrins, K. Sivakami Sundari
PG Scholar, Chandy College of Engineering, Chandy Nagar, Tuticorin-5 Tamil Nadu, India Professor, Chandy College of Engineering Chandy Nagar, Tuticorin-5 Tamil Nadu, India
Abstract-Identification of Retinal Blood Vessel is much required for detecting vision capacity, Diabetic retinopathy, and also cardiovascular diseases like ophthalmic pathologies, hypertension. This work examines the blood vessel segmentation methodology in two dimensional retinal images acquired from a fundus camera. Many deduction methods are available but the results are not satisfactory. The measurement of the various morphology parameters of these retinal blood vessels plays a vital role for detecting the diseases and obviously clear that the wrong identification of these measurements leads to a wrong clinical diagnosis. Hence there comes the need for identifying the true vessels of the segmented retinal images. The proposed project work presents a new method for identifying the true vessels with apparent pre processing phase: as a first step the acquired images are pre-processed with three steps as i. Gray scale conversion, ii. Detection of edges and corners, iii. Median filtering for the removal of any salt and pepper noise if present, and then a Gaussian model for segmentation. True vessels are identified based on the connectivity of the eight neighbourhoods and also on the black to non black transition in the morphological opening. In the second step the pre processed and then segmented image is used for the identification of true vessels by modelling the segmented graph though finding the morphological parameters. Locating the cross over’s and the bifurcations are slightly the hard-hitting effort and done in various ways. The work was carried out with the images from the publicly available DRIVE and STARE databases which are widely used for this purpose, since they contain retinal images where the vascular structure has been precisely marked. Algorithm is simulated in Matlab and the results suggest that our proposed system can be efficiently used to identify the true blood vessels,
  View Paper