Researches on Medical Image Registration Algorithms Based on Mutual Information
|School||Beijing Jiaotong University|
|Course||Pattern Recognition and Intelligent Systems|
|Keywords||mutual information Powell algorithm improved genetic algorithm hybrid optimization algorithm|
Medical image registration, an important research subject consisting of multiple disciplines involves information processing, computer technology, modern medicine and so on. It plays an important role in clinical diagnosis, treatment and preoperative plan. With the continuous development of medical image registration technology, many image registration methods has been proposed. However, image registration method based on mutual information not only does not require pre-process of image but also has high precision and strong robustness. In recent years, the method has caught many experts’ eyes.In this paper, we concentrate on the medical image registration algorithm based on mutual information. Through reading a large number of literatures, we seriously analyzed the research background and the current development situation of medical image registration. At the same time, we study the theory knowledge of image registration methods further and detailedly analyze the algorithm principle as well as the advantages and disadvantages of the algorithm.We improve the crossoveroperation of genetic algorithm considering the shortings of the standard genetic algorithm, such as slow convergence and prematurity that will result in mismatching. Image registration algorithm based on mutual information has high complexity and slow speed. To solve the problem, we combine the improved genetic algorithm and Powell algorithm. The method makes full use of the global search capability of genetic algorithm and the local search capability of Powellalgorithm. Meanwhile, the method overcomes the blindness of the initial point selection of Powell algorithm. In a word, the method we proposed greatly shortens the image registration time and improvesapplication value of the medical image registration in the field of clinical medicine.We conduct simulation experiment through Matlab. The experimental data show that the algorithm we put forward is effective.