Virtual Chinese Human (Male No.1) Multimodality Image Registration Research

Digital virtual human is a frontier of research combined by Medicine, Information Technology and Virtual reality technology. Normally, there are three methods to obtain the research materials: CT, MRI and Mechanical cutting. And the Data sets of three modalities were obtained by above methods. This Data set takes new mass omni-directional materials for human research. How to take full advantage of these Data is also a question worth considering by most researchers. Information fusion of non-homologous pictures is a effective way to solve this problem. If non-homologous information can be comprehensively analyzed, and information from various kinds of images, which reflect the digital human body, can be fused, then the information expression ability of digital human body will exceed the sum of expression abilities by every single image.Based on the thought of multimodality images complementary information, integrated interpretation, we use multimodality image registration technique to resolve registration of the three images modality of Virtual Chinese Human, and realize the registration of above three modality images in two-dimensional space.In this paper, according to the features of the above-mentioned three images, CT image was chosen as the reference image, when MRI images were registered, the best match parameter was searched by discriminating the largest mutual information between the two kinds of images gradient; while color slice images were registered, the optimal spatial transformation parameter was obtained by registration based on the anatomical structure feature; then, images needed registration were transformed on the basis of the optimal spatial transformation parameters.Three images of different modality data were registered, and compared with high-quality manual segmentation image datasets, registration accurate rate is 95.8%. The results show that registration of three images can be accomplished exactly by this methods, it provides reference for multimodality image registration of digital human.