Algorithm Reseach On 3D Cerebrovascular Images'centerline Extraction

In the clinical diagnosis and therapy of cerebrovascular disease,computer processing on 3D cerebrovascular images supplys doctors a tool to observe 3D cerebrovascular structures at any angle,and can be useful in the diagosis and therapy.Because blood vesssels brances is complex and its shape is thin,how to get the the exact description of cerebrovascular structures has been intractability.The frame of 3D cerebrovascular structures depand on the its centerline,which has the same topology and geometry shape with the cerebrovascular images.Base on methods for 3D centerline extraction of objects which have been introduced in recent years,in this paper,algorithm reseach on 3D cerebrovascular images'centerline extraction is presented.The main parts of the study of this thesis are as following:1) First, we make a summary of the methods for 3D centerline extraction of objects which have been introduced in recent years.2) In this part, a centerline extraction based on 3D Euclidean distance transform is presented.This method include the 3D local maxima selection, neighbour maxima selection and saddle points selection.This method is automatic.3) In this part, a centerline extraction based on Hessian matrix is presented.This method include 3D Euclidean distance transform, computer the Hessian matrix in every voxel,and visibility test.This method is also automatic.4) Most methods for 3D centerline extraction of objects which have been introduced in recent years get only the centerline points.Our objective is that we can get 3D cerebrovascular structures clearly, so how to get the path of centerline is very important.We get the path with all the points by Dijkstra's shortest path algorithm.