Study Of Medical Image Registration Methods Based On Multi-features And Artificial Immune Algorithm

With the development of science and technology, medical image has become an important component for clinical diagnosis, and medical imaging technology has also become the most active areas of the research of the medical diagnosis. Medical image registration technique is one of the basic questions.Image registration is the process of space alignment for more than two or two images taken at different times, different angles or different devices. For medical images, image registration can be divided into single-mode medical image registration and multi-mode medical image registration. Single-mode medical image registration is mainly used for monitoring the development of disease. As different mode medical images reflect different information, we could registrate them and fuse them to one image to provide a more comprehensive and accurate information for clinical diagnosis and surgical treatment.The research background, common methods and translation models of image registration are introduced firstly in this thesis, then the major registration algorithms and categories are described. At present, there are so many methods of image registration, but neither of them can be fit for both the single-mode and multi-mode images. The algorithm only based on the points feature costs short time but with low accuracy, and it's less effective for the multi-mode medical image registration, and the algorithm based on gray feature only is accurate, but it will cost long time especially for single-mode image registration. In order to sum up the advantages of both the methods, points feature and gray feature are combined together in this paper. To address this issue, the algorithm based on points feature combined with the gray feature using the artificial immune optimization algorithm for parameter optimization is choosed in this article.In this paper, Harris operator is used to extract the feature points, and is improved to solve the higher rate of false extracting. Mutual information based on gray feature is used for similarity measure. For searching parameters, artificial immune optimization algorithm is introduced to the medical image registration in this paper to avoid long time of the traditional algorithm and easily geting into local extremum of the genetic algorithm. Lots of registration experiments have been done in this paper. Four different registration algorithms are compared, such as the improved Harris operator combined with the gray feature, the classical Harris operator combined with the gray feature, single point feature and signal gray feature, and the results show that the algotithm of improved Harris operator combined with the gray feature could get the best feature extract results. The algorithm proposed in this paper and the traditional algorithms are applied to the single-mode medical image registration and the multi-mode medical image, and compare the experimental results using artificial immune optimization algorithm, the traditional Powell optimization algorithm and genetic algorithm repectively. The experiments results show that the algorithm proposed in this paper is accurate, effective and robust.