An automatic registration approach for whole body PET images and head MR images

Registration of multimodality images is currently a hot topic in medical image processing of which the whole body PET images and head MR images automatic registration is one important application. In this application, the great differences of image scanning ranges make the direct application of mutual information registration method failure. This paper presents a hybrid 3D registration approach based on normalized mutual information which can match the whole body PET images and head MR images automatically. The performance of the hybrid approach was evaluated for rigid-body registration of MR and PET images of the same patient and the results show the hybrid approach can be satisfactory by doctor. © 2011 IEEE.