Multi-Feature Based Bayesian Segmentation of Cerebrovascular from Time-of-flight Magnetic Resonance Angiography  

作　　者：HAO Jutao LI Minglu 

机构地区：[1]Department of Computer Science and Engineering, Shanghai Jiaotong University, Shanghai 200240, China

出　　处：《Chinese Journal of Electronics》2008年第2期252-256,共5页电子学报（英文版）

基　　金：This paper is supported by the National Grand Fundamental Research 973 Program of China （No.2006CB303000） and the Natural Science Foundation of Shanghai （No.05ZR14081）.

摘　　要：Vessel analysis in medical images is important both for diagnostic and intervention planning purposes, especially, a three-dimensional representation of vasculature can be extremely important in image-guided neurosurgery and pre-surgical planning. In this paper, a Bayesian approach is proposed to aggregating geometric shape and intensity features for whole cerebrovascular tree extraction from Time-of-flight Magnetic resonance angiography （TOF-MRA）, while most of the current segmentation methods solely deal with the latter. In this method, we first utilige scale space analysis to get shape feature of blood vessels, then both shape and speed features are incorporated into a Bayesian segmentation framework. Maximum a posterior （MAP） method is used to estimate the posterior probabilities of vessel and background for classi- fication. The experimental results show that the proposed method can obtain a better quality of segmentation than those sole feature utilized methods.

关 键 词：Bayesian segmentation Markov random field （MRF） Scale space analysis Maximum a posterior （MAP） estimation. 

分 类 号：R445.2[医药卫生—影像医学与核医学]