基于图像特征的CT-MR图像配准  

CT MRI Registration Based on Images' Geometrical Features

作　　者：赵富强[1] 于红玉[1] 马斌荣[1] 李坤成[2] 

机构地区：[1]首都医科大学,北京100054 [2]首都医科大学附属宣武医院,北京100053

出　　处：《北京生物医学工程》2000年第4期206-211,共6页Beijing Biomedical Engineering

摘　　要：本文采用了基于图像特征的相关性算法对 3- D CT- MR图像进行配准。首先使用基于局部坐标系的二阶导数算子对 CT和 MR图像进行处理 ,获取由图像的“脊”或“谷”组成的特征图像 ,然后在进行刚体变换条件下对特征图像进行相关计算 ,使其相关系数达到最大 ,从而获取配准后的图像。该方法不需在成像过程中人为增加标记物 ,后续处理中不需人为干预 ,是一种自动的图像配准方法。In this paper is described an automated approach to register CT and MRI brain images based on their geometrical features. Local second order differential operators in scale space are applied to each type image date, so as to produce feature images, depicting ridges or vallegs. Then correlation calculation is carried out for feature images under the condition of rigidity, so that the coeffiuent of correlation is maximum. The resulted CT with MRI feature images are then obtained. This method does not need human interaction and interactive operation is also not needed during post processing. Results obtained using this method are presented. It is proved that this method is practically useful.

关 键 词：配准 图像特征 相关性 二阶导数 CT-MR图像 

分 类 号：R814.42[医药卫生—影像医学与核医学]