Artikel
Enhanced segmentation and visualization of blood vessels based on CT-angiography
Fortgeschrittene Segmentierung und Visualisierung von Blutgefässen basierend auf der CT-Angiographie
Suche in Medline nach
Autoren
Veröffentlicht: | 4. Mai 2005 |
---|
Gliederung
Text
Objective
Current CT scanners provide high resolution three-dimensional (3D) data. Despite the improved accuracy of scanned data, a reliable visualization of blood vessels in the immediate vicinity of bony structures like the scull base is still a challenge in many situations. The authors present a simple and efficient segmentation algorithm providing a fast and accurate insight of blood vessels and bone structures.
Methods
The segmentation pipeline is composed of (1) threshold segmentation, (2) distance field computation, (3) path search and (4) false connection removal. The scanned volume is binary segmented into bone and blood vessels on the one hand, and soft tissue and air on the other. The binary volume is processed to compute a distance field. This results in a value for each voxel corresponding to the distance from the given voxel to the closest boundary. Within the distance field volume, a path is searched that maximizes the distance to the boundaries. Finally, oblique cross-sections are computed along the obtained path. Those points where a minimum cross-section is found are identified as most likely false connection points and therefore removed.
Results
After segmentation of blood vessels and bone structures, they can be visualized in 3D using indirect volume rendering. The compound result can be directly displayed on the screen, or exported to a file in several standard 3D formats for further use during diagnostic or intraoperative procedures. The results show that false connections between vessel and bone structures are satisfactorily removed, thus providing a more accurate and reliable 3D data representation.
Conclusions
A new solution for optimised visualization of blood vessels in CT angiography is presented. The proposed method consists in a simple segmentation algorithm, which combined with indirect volume rendering allows an efficient analysis of CT angiography data in a 3D environment.