Research On Liver Non-rigid Registration Methods Based On Abdominal CT Images

With the development of modern computer technology and digital medical equipment, digital image in medicine has been one of the most important methods in disease diagnosis for clinicians and experts. Medical image registration has important guiding significance for CAD (Computer Aided Diagnosis). Registration accuracy impacts the quality of feature extraction and diagnosis of the CAD system directly. Therefore, improving the accuracy of image registration has important application value for clinical medical research.We propose a new medical image registration method based on hybrid information model combined with gray-scale information and feature information based on medical image registration theory. Firstly we complete the rigid image registration based on gray-scale information. Then, we make the further non-rigid registration based on feature information. We need to extract feature points and contours of the images in elastic registration based on feature information. In this paper, it uses the improved SEMISURF algorithm for feature point extraction to resolve the limitations of SURF algorithm. Then, it uses wavelet transform to extract the contour and optimizing processes the contour information which would obtain the valid feature information of images. Finally, we complete the elastic image registration based on the thin-plate spline algorithm for the images with double feature information. During the parameter optimization, it proposes a hybrid optimization algorithm based on RMI-SAPSO to resolve shortcomings of traditional particle swarm optimization algorithm. It chooses regional mutual information as objective function firstly, and then uses SAPSO to optimize the regional mutual information. The proposed algorithm increases the extraction of the spatial information through introducing the regional mutual information and improves the registration accuracy of the local area of the image.Applying our methods to liver registration experiments of the practical abdominal CT images, the experiments results show that the proposed method are superior to traditional methods in terms of both the accuracy and efficiency of the registration. The SEMISURF algorithm can effectively extract feature points of the deformation complex images. The optimization algorithm based on RMI-SAPSO can optimize the parameters ideally. The hybrid information model can be more accurate for medical image registration and get better elastic registration results.