基于最大互信息的人脑多模图像快速配准算法

Fast Algorithm of Brain Multi-modality Image Registration Based on Maximal Mutual Information

对脑图谱开发过程中来源于不同成像设备的多模图像进行配准。对预处理后的数码图像和MRI图像,首先提取图像的轮廓,采用基于轮廓的力矩主轴法计算初始平移量和旋转量,然后设定初始缩放系数,将此初始配准参数作为改进单纯形法的初始参数,以互信息作为相似性测度迭代搜索,使互信息最大,从而实现最佳配准。结果表明本算法不需要人为预调整待配准图像的分辨率,自动化程度高,配准速度快,精度较高,能够满足脑图谱开发过程中的多模图像配准要求。

To perform the multi-modality image registration from the different imaging devices during the research of human brain arias. After distilled the contour of pre-processed digital image and MR image, the initial translation parmeter and rotation parameter by using the method of principle axes based on image contour were calculated. Then the initial zoom parameter and search for the best matching parameters were set to make the mutual information maximal, using the simplex method with the mutual information as the comparability criterion. Result shows that this algorithm doesn＇ t need manual pre-adjustment of image resolution. So it has high degree of automation and the advantage of high registration speed and high registration accuracy. It can meet the demand of multi-modadity image registration well during the research.of human brain atlas.