During surgery and radiation therapy,demands for higher quality of quantification and visualization of cerebral vessels need to be met.Hence,automatic (or semi-automatic) extractions of accurate cerebral vessel information seem particularly important.This paper reviews the cerebral vessel segmentation methods used in recent years,including statistics-based segmentation,deformable model,multi-scale method,prior knowledgebased method,and so on.The methods are divided into the following three categories:voxel intensity based approach,vascular tubular structure based approach,and prior knowledge-based approach.