A Novel Method for 3D-Segmentation of Vascular Images

Abstract

Constructing the accurate digital model of vessel networks is critical to vascular tissue engineering, in which the segmentation of vessel plays an important role. However, the existing segmentation methods are not able to achieve the goal of accurate segmentation of vessel networks. This paper presents the development of a method for vessel segmentation based on a data structure of octree and 3D region growing. Firstly, the volume data of vessel images are divided into different data groups according to the predetermined depth value of octree, and then the optimal slices sequence is defined by analyzing the octree’s nodes which contain the vessel region. Then, the vessel segmentation is conducted from the vessels images of octree nodes based on 3D region growing. Finally, the treated data blocks are reset and the segmentation results of the whole volume data are obtained. By applying this method to the volume data of vascular images from MRA, accurate vessel segmentation results are achieved. This work would represent a significant advance for digital modeling of vessel networks. Key Words: Vascular tissue engineering; Image segmentation; Region growing; Octree