Extraction of blood vessels using multilevel thresholding with color coding

Abstract

The proposed work deals with the design of algorithm for segmentation of blood vessels in order to extract the geometrical features of analyzed vessels. The algorithm was tested on a sample of 25 clinical data to identify blood vessels direction and suppress surrounding tissues. The proposed methodology is effective for practical purposes, because identifies minor visual differences on the surface of blood vessels. These changes often represent pathological lesions that cannot be identified from the native data. The algorithm is based on the identification of several significant levels to make ideal thresholding, according to object features. For even better diagnostic results image background is subsequently filtered out from the segmented image, background may have a disturbing effect in the diagnosis and the final segmented image contains only analyzed blood vessels.