基于互信息快速算法的多模医学图像配准

Multi-Modality Medical Image Registration Based on Rapid Algorithm of Mutual Information

互信息作为相似性测度在多模医学图像配准领域得到了广泛应用,具有高精度和稳健的特点.但频繁的互信息计算降低了配准效率,同时对两幅图像重叠区域比较敏感.采用归一化互信息为测度,降低互信息计算中灰度级的快速算法,通过多模CT/MRI配准实验证明,可以在确保配准精度的同时缩短时间提高配准效率.

The similarity metric mutual information with high precision and stability has been used widely in multi-modality medical image registration,but frequent calculation of mutual information reduces efficiency of registration.Moreover,it is sensitive to the overlap of two images.The experiment with CT/MRI image registration uses normalized mutual information as a metric and decreased the grayscale of mutual information＇s calculation.The rapid algorithm proves that it also improves efficiency of registration under a certain registration precision.