Mutual Information Based Multimodality Medical Image Registration Research

Medical imaging has been an indispensable part of modern medical treatment. The medical images from different modalities can complement each other. The useful image information often need to be composed in order to give more comprehensive information by using more than one modality. The first step of composing is to image registration, the next step is image fusion. Image registration research is the emphases of this dissertation.Mutual information registration method thinks that the mutual information of the wrapped part of the images reaches the maximum when two images register best. Compared to conventional registration methods, the predominance of Mutual information registration method is that it doesn't need to suppose the relation of the images, and that it need no image segmentation and image preprocess, and that it can bring to success automatically without alternation with people.The intensity of image pixel can be thought of as a stochastic variable. Thus, the mutual information of the two images can be computed by joint probability distribution and marginal distribution theory and entropy theory. The joint and marginal intensity histograms of the wrapped part of the images don't need the computation of the derivative of mutual information, so them can be used to compute joint probability distribution of the two images in order to reduce realization difficulty of the method. Because the pixel is likely to not be integer after space transformation, interposition technology is needed. The double-linear PV interposition we select won't produce new intensity, so it is better than other interposition methods. The progress of computing maximal mutual information can be looked upon as a multivariable optimization. Powell optimization method we select is both fast and exact. The method can search maximum of the mutual information successfully and realize image registration..From the result of the experiment, I believe that the maximal mutual information Pewell optimization image method we research is a exact, robust and completely automatic multimodality medical image registration method.