Abstract: In order to realize the automatic registration of thorax multimodal medical images，a rapid image registration method based on hierarchical adaptive free-form deformation is proposed.Firstly，the edge of thorax is extracted by C-V level sets algorithm，and feature points are matched automatically which is based on parallel computing method.Then，the global coarse registration of thorax multimodal medical images is modeled by principal axes algorithm.Finally，the local fine registration of thorax multimodal medical images is realized by a Free-Form Deformation（FFD） based on hierarchical B-splines.Moreover，the FFD parameters are generated rapidly by gradient descent and maximization of mutual information.Experiments demonstrate the good performance of the proposed method.

Key words: adaptive free-form deformation, multimodal image registration, hierarchical B-spline, gradient descent, parallel computing, level sets method

摘要： 为了实现胸部多模态医学图像的自动配准，提出了一种基于层次B样条自适应自由变形的快速配准方法。首先采用C-V水平集方法实现感兴趣区域的提取，并基于并行计算实现自动配对特征点；接着，采用矩主轴法对多模医学图像进行全局粗配准；最后，基于层次B样条自适应自由变形法对多模态医学图像进行自动细配准，并且采用梯度下降法以及最大信息熵准则加速求自由变形系数。实验证明该方法不仅效率高，而且配准效果好。

关键词: 自适应自由变形法, 多模图像配准, 层次B样条, 梯度下降, 并行计算, 水平集方法