基于人工免疫及最大互信息的胸部多模图像配准  

Thorax multimodal image registration based on artificial immune algorithm and maximization of mutual information

作　　者：李彬[1] 欧陕兴[1] 田联房[2] 毛宗源[2] 

机构地区：[1]广州军区广州总医院放射科,广州510010 [2]华南理工大学自动化科学与工程学院,广州510640

出　　处：《计算机应用研究》2009年第5期1951-1954,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(30570458);广东省自然科学基金资助项目(8451064101000631)

摘　　要：为了实现胸部医学图像的自动配准,提出了一种基于人工免疫及最大互信息的配准方法。首先定义了最优配准的目标函数,接着运用人工免疫算法,结合最大互信息熵函数产生最优的仿射变换系数,从而实现医学图像自动配准。该人工免疫算法中,抗原是指最大互信息熵目标函数,而抗体是指最优的仿射变换系数。实验证明该方法配准效果较好。In order to realize the automatic registration of pulmonary medical images, this paper proposed an image registration method based on artificial immune algorithm and maximization of mutual information. Firstly, presented the objective function for the optimal registration. Then ,presented an artificial immune approach to automatically generate the optimal parameters of affine transform, thus realized the multimodal image registration. In this approach, regarded the objective function as antigens, and regarded the optimal parameters of affine transform as antibodies. Experiments demonstrate the good performance of the proposed method.

关 键 词：人工免疫算法 多模图像配准 最大互信息 最优化 胸部 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]