基于互信息的多模医学图像配准

以互信息配准原理为基础，讨论了互信息配准算法应用于图像的几何对准并给出了初步的评估结果．通过CT／MR、PET／MR配准后的结果与现有的标准进行比较，说明互信息法不需要对图像进行预处理，可以实现多模医学图像配准，并达到亚像素级的配准精度，以满足临床应用的要求．该算法可靠性高，对图像几何失真、灰度不均匀性及数据的缺失等不敏感，并适当调整初始化参数，可以避免明显的误配准．

Medical Image Registration Based on Mutual Information

Based on the principle of registration of mutual information, this paper discusses that the mutual information algorithm can be used in the geometric alignment of image registration and gives a preliminary result of assessment. By comparison of the results after image registration of CT/MR, PET/MR and available standard, it demonstrates that multimodal medical image registration and sub--pixel registration accuracy can be achieved using the mutual information without any preprocessing steps, which makes this method very well suited for clinical applications. The method is highly reliable and not sensitive to geometrical distortion, intensity inhomogeneity and data missing, and it can avoid obvious errors by adjusting initial parameters properly.