利用边缘和表面特征实现脑图像的自动配准

本文中我们使用基于CT、MR和PET图像等值特征表面的配准算法对多模医学图像进行了配准研究，在CT、MR和PET的原始图像中提取等值特征表面，进行图像的几何对准，并对结果进行初步评估，同时对该算法的稳健性，搜索最近点策略和采样策略进行了研究，结果表明；这种方法能够达到亚像素级的配准精度，是一种稳健、高精度、全自动的配准方法。

Automated registration of brian images using edge and surface features

An automated registration method based on the equivalent feature was discussed in this article.The skin surface of the head,the outer brain surface and the inner skull surface are commonly used as the corresponding surface for image registra-tion.Preliminary evaluation result shows that Sub-voxel accuracy can be achieved with this approach.We also investigate the performance of the method of searching closet points and sampling action.The algorithm we presented is a very robust and au-tomated multi-modality registration method with high accuracy.