基于最大互信息的人脑MR-PET图像配准方法

利用最大互信息法进行多模医学图像配准近来成为医学图像处理领域的热点。MR和PET图像配准对研究神经组织的结构关系和引导神经外科手术有着重要的指导意义。本文描述了一种基于互信息的人脑MR－PET图像配准方法。我们将这种方法应用于图像的几何对准并给出了初步的评估结果。由于不需要对不同成像模式下的图像灰度间的关系作任何假设，最大互信息法是一种稳健性强，可广泛应用于基于体素的多模图像的配准方法。

Method of Human Brain MR-PET Image Registration Using Maximal Mutual Information

There has recently been considerable interest in medical image processing community in the use of maximal mutual information for automatic registration of images from different modalities. Applyication of registration methods to MR and PET images is very important in neurology to study structure function relationships and to guide neurosurgery. In this paper is described an automated mutual information based method for human brain MR PET image registration. We have applied this method to the problem of geometric alignment of autoradiographs. Preliminary evaluation results are presented below based on our experiments. Because no conditions are required regarding the nature of the relation between image intensities of both modalities, Maximal mutual information based method is very robust and may be quite useful in multimodalities voxel based registration.