Centerline-based Method For Automatic Segmentation Of Cerebral Blood Vessels

Vessel centerline extraction and vessel segmentation play important roles in the quantitative analysis,auxiliary diagnosis,lesion location,and planning treatment of vascular diseases.The results of centerline extraction and segmentation provide important information for the diagnosis and treatment of vascular diseases.However,the existing centerline extraction and vessel segmentation methods need to be further improved in integrity and accuracy.Based on the tradition shortest path method,this paper proposes a multi-information fusion centerline extraction method for centerline extraction of annular vessels.At the same time,in order to segment the blood vessels automatically,an automatic vessel segmentation method based on the centerline and graph cut is proposed.The specific research work is as follows:1)A centerline extraction method based on multi-information fusion for ring-like blood vessels.The results of the centerline extraction have important implications for the quantitative description of different features of vascular structures.Firstly,according to the result of vessel segmentation,the shortest path method is used to obtain multiple original centerlines based on multiple random seed points.Secondly,the topology information of original centerlines is integrated to ensure the integrity and continuity of the centerline.And the center voxels are reserved based on the maximum boundary distance field.Finally,the redundant point judgment criterion is used to keep the centerline with single voxel width.We conducted experimental verification on simulation data and brain data.Compared with the traditional method,the accuracy and integrity of the centerline have been significantly improved.2)Segmentation of blood vessels based on centerline and graph cut.The results of vessel segmentation are commonly used in vascular analysis such as vessel matching,three-dimensional reconstruction,and motion estimation.The traditional graph cut method transforms the segmentation into the energy minimization problem,which relies on the artificial marking of the target and the background.This study uses the centerline instead of artificial marking and proposes an automatic vessel segmentation method based on graph cut.First,the study adopts vesselness filtering to enhance the original blood vessel;then topological thinning is used to extract the center line of the blood vessel.Second,the centerline is used to get the information of initial target and background.We set the probability of centerline points,and then calculate the probability of the voxels outside the centerline,which determines the weight of edge.Finally,vessel segmentation was performed using the graph cut.We use this method and two comparison methods to verify the validity on the simulation data and brain data.The results show that this method realizes the automatic segmentation of blood vessels on the premise of guaranteeing the accuracy,and improves the efficiency.