A Novel Subsampling Method for 3D Multimodality Medical Image Registration Based on Mutual Information

Mutual information (MI) is a widely used similarity metric for multimodality image registration. However, it involves an extremely high computational time especially when it is applied to volume images. Moreover, its robustness is affected by existence of local maxima. The multi-resolution pyramid approaches have been proposed to speed up the registration process and increase the accuracy of the result. In this paper, we present a new improved method of sample selection for multi-stage registration based on mutual information. Instead of down-sampling of the whole image as it is done in the pyramid methods, we propose a new technique to find a suitable subset of image samples based on image information content, which results in a better estimate of the optimal transformation. A comparison for MR images indicates that our proposed method yields better registration than subsampling method, especially when subsampling factor is low. The experimental results involving three-dimensional clinical images of CT, MR and PET are presented for rigid registration.