Multistage CT/PET Medical Image Registration Based on Gradient and Mutual Information
Ong L · Microcomputer Information · 2008
Medical image registration based on mutual information is researched in this paper. As the absence of space information, it is easy to get into local maximum and results in error registration, meanwhile, it needs long time as the computed quantity of mutual information is large. A multi-stage registration approach based on mutual information (MI) and gradient information is proposed in this paper and applied in the registration of CT/PET. Morphology method is used to reduce noise of the PET image, contrast-stretching and histogram equalization are used in order to get distinct gradient image. Firstly, principal axes method is used to the gradient images and then is the mutual information. The results show that the multi-stage registration approach is obviously better than the traditional mutual information in the registration of CT/PET.