Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

3D Topological Model Construction of Virtual Hepatic Vascular Tree and Its Application

Author WangPeng
Tutor FangBin
School Chongqing University
Course Applied Computer Technology
Keywords hepatic vasculature thinning branch-points identification trie tree graph representation
CLC TP391.41
Type Master's thesis
Year 2013
Downloads 3
Quotes 0
Download Dissertation

The three-dimensional reconstruction of the liver images and to establish a virtualliver model is a way to compensate for the weakness of the two-dimensional imagesevaluation. The structure of vasculature describes the anatomic location and hierarchyof vessels, also embodies the blood supply.In morphology, the centerline is way to describe the shape of the object, whichreduces the dimensionality, combines the contour of the object and its area information,and represents the clue in visualization. We can transform linear connected structure ofcenterline into abstract form of tree or graph for characteristics matching. Therefore, theresearch on the representation and recognition of the target based on the centerline isimportant, which is in the field of pattern recognition and computer vision. The coretechnologies in this research are centerline extraction technology and objectrepresentation based on the centerline. For the former, a large number of centerlineextraction algorithms have been proposed, while the research for the latter is still verylimited. Directly extracting structural characteristics of the target object from thecenterline images is difficult and inefficient.This paper presents a method for generation of the graph representation of vasculartree. Firstly the thinning and single-voxel operations are executed on the hepaticvascular tree through template matching, and by analyzing voxel adjacent relationshipthe branch-points are identified. An algorithm based on three-dimensional connectedcomponent labeling and breadth-first search is proposed for cycle removing, and theinformation of vessel diameter and length in three-dimensional images is used forpruning. As a result, the centerline which is coincident with the actual situation of theliver vascular tree is extracted. Then the centerline is traversed and the graphrepresentation of vascular tree is built with a trie tree at the same time. The statisticalresults showed that the extract centerline obtained through this method is connected andof high accuracy, and can be applied to the measurement of the vessel length anddiameter and vascular grading. After many experiments, the graph representation canabstractly represent the vascular tree well using the data structure of trie tree.The diameter of the hepatic portal vein and hepatic vein can be used as a basis fordiagnosis of liver disease which is of great significance. In this paper, the center line ofthe vascular tree and vascular tree itself are partitioned using its topological model, and then calculation of length and diameter of the vessel branches is preformed using theEuclidean distance. Strahler method are used to assign the levels to branches in order tomeet the needs of the surgeons in diagnosis using different filter parameters, and theresults of grading is agreed with the vascular tree hierarchical relationship.Bland-Altman analysis proves that there is a high degree of accuracy of this method inthe calculation of the length and diameter of the vessel.These algorithm steps have been applied in the virtual liver surgery planningsystem and works well, and we can calculate some parameters such as vessel length,volume and diameter, and partition the liver interactively, and with these surgeons areable to analyze the vasculature quantitatively.

Related Dissertations
More Dissertations