Segmentation of Blood-Vessel Tree in Whole-Body MRI Data

Abstract

The aim of this paper is to introduce a method of blood-vessel tree segmentation in 3D volume and the visualization of the processed data. The method is based on the detection of tubular structures which the blood-vessels resemble. The processing is based on the information that is carried in the derivatives of the original volume. Second derivatives of each voxel are computed and organized into a Hessian matrix. The used method is based on the calculation of the eigenvalues of this matrix for each voxel. Those eigenvalues carry the information of the structure shape. Based on this information, a new parametric volume is created that represents the ’vesselness’of each voxel. The processed volume is visualized with Visualization ToolKit.