Modeling and Features Extraction of Blood Vessels based on Soft Regional Segmentation

Abstract Analysis and visualization of the blood vessels has a crucial impact for the clinical practice. One of the most important aspects of such analysis is the blood vessels features extraction characterizing their state and manifestation. Such procedure may be done by using the mathematical modeling of the blood vessels which is not conventionally possible from the native records. We have proposed a mathematical segmentation model for extraction and modeling of the blood vessels structure from the CT angiography native records with a target of differentiation of the physiological blood vessels from calcification spots indicating areas where the blood vessel is damaged. Such model consequently allows for objective quantification of the calcification amount, as an important clinical parameter of the damage level of the blood vessel. The method is based on the histogram partitioning, and consequent classification via predefined number of the fuzzy triangular classes representing individual parts of the blood vessels. Consequent part of the segmentation model takes into account spatial relations inside of each region to make a model accurate, and robust against the noise and artefacts. As a part of our analysis we have tested the computation complexity of the proposed segmentation model.