Research On Segmentation Of Coronary Artery From Msct Data Based On Level Set Method

Because of high incidence, high death rate and other characteristics, coronaryartery disease has been a common disease which is a threat to human health theseyears. X-ray coronary angiography technology (CAG) and multislice computedtomography technology (MSCT) play a key role in the diagnosis of coronary arterydisease. Because of its isotropic scanning, non-invasive imaging, high spatial andtemporal resolution and other advantages, the multislice computed tomographytechnology is widely used in coronary artery disease analysis and diagnosis, andachieves good effect. In this article, in order to obtain higher accuracy ofsegmentation, basing on sufficient data compilation and research on desk, we useimage segmentation model based on level set method to solve the difficulty ofaccessing coronary artery information.Taking into account the division of high precision and topology adaptabilityadvantages, the level set method has obtained wide range of applications after it hasbeen introduced into the field of image segmentation. Level set method is a kind ofmodel-based segmentation methods in image segmentation in recent years. In thispaper, we use Chan-Vese model for image processing as its high precision andtopology and good adaptability. For the model can not segment uneven gray imagesand the selection of the initial contour has dependency, this thesis discusses animproved image segmentation model with a fusion of fuzzy clustering, uses theinformation of fuzzy clustering to guide the selection of initial contour in Chan-Vesemodel. First, deal the CT data to gain clustering information and membership matrixusing fuzzy clustering method; second, define the initial contour of level set methodusing the clustering information; last, extract the coronary artery using C-V model andcomplete the image segmentation. Then this paper studies3D medical imagereconstruction algorithms and completes the reconstruction of coronary artery usesurface rendering and volume rendering method, which aims at providing doctorswith real sensory effects. The results are used to aid diagnosis of coronary heartdisease.