A hybrid optimization method for maximum mutual information registration

Maximum mutual information (MI) is widely used in multimodality registration. However, traditional registration based on MI is difficult to be adopted in clinical routine for its time-consuming. We proposed a hybrid optimization method which can reduce time for one evaluation of the MI criterion by using incremental coordinate transformation method and also can reduce the whole iteration time by using multi-resolution as the search strategy. The performance of the hybrid optimization method was evaluated for rigid-body registration of CT and PET images of the same patient, and the results showed the hybrid method was faster than traditional maximum mutual information method.