The Research On Segmentation Algorithm Of Intracranial Vessel

Researchers have indicated that accurate segmentation data of intracranial vesses are very important to fluid mechanic simulation. Even one tiny mistake will possibly result to a big difference, so how to segment vessels effectively is a key to the process of mechanic simulation. Medical imaging instruments are vulnerable to both internal and external factors, making images featured blurred and low contrast. Further more, intracranial vessel is so small that it could be much difficult to segment vessel images.Active contour model(ACM) is one of hottest image segmentation model, which have some advantages over others, for example, it is able to resist noise and process the image with intensity inhomogeneous. This paper focuses on proposing some improved segmentation models and applying them to segmenting 2D and 3D intracranial vessel data based on ACM. Main jobs are as below:1.This paper proposes a kind of SBGFRLS model combined with local threshold model,this model applies local threshold to level set method, which enhances the ability to segment images with noise and intensity imhomogeneous.2.Based on characters of intracranial vessels, it proposes a kind of ACM model by introducing the difference image. The test indicates that this model has multiple segmentation characters, which can not only segment images with noise but also with intensity inhomogeneous, in addition, it has a quality of high segmentation efficiency.3.Aimed to segment 3D tomography images, this paper improves the CV model and SBGFRLS model and then applies them to 3D volume data. The test shows that the improved models can obtain smooth 3D vessels by rebuilding the 3D segmentation data.4.We develop a 3D rebuild software(3DRebuilderTool) based on MITK engine in line with our latter work of intracranial vascular fluid mechanics simulation, which is able to rebuild and show 3D graphics in computer according to 3D data. In addition, it has other functions such as segmentation, virtual cutting, measurement, and exporting 3D mesh files which are data source for fluid mechanic simulation of intracranial vessels.