A Mixed-Type Registration Approach in Medical Image Processing

Medical image registration is a critical step in medical image processing. In this paper, a mixed-type image registration approach is presented, which combines the segmentation-based and voxel-based registration. Firstly, the experimental images are preprocessed, including digital imaging and communication of medicine (DICOM) format conversion, denoising, and segmentation. Then mutual information (MI) as a similarity measure is used when the two images matched, and finally the optimal image transform is chosen by using optimization strategies. Experimental results show that the novel approach has low computational complexity, fast speed, and high accuracy. Moreover, it can fully take advantage of both registration approaches based on segmentation and voxel similarity, which improves the accuracy and speed of the registration effectively.