Research Of Multi-Resolution2D-3D Medical Image Registration Based On Mutual Information And Ant Colony Algorithm

The2D-3D medical image registration which is a part of the medical image registration was considered one of the research hotspots currently. And the most typical application is to intervene in surgical navigation. The2D medical image is lack of spatial information comparing with the3D medical images, so it can not provide enough information for clinical diagnosis and treatment. In order to compensate for this part of the spatial information and assist clinical treatment, the image registration technique was applied to registering the2D medical images which were collecting in the operation with the3D information that was obtained before. With the increasing emphasis on personal safety, the improving accuracy and efficiency of image registration is means the improving of security. So we study to find out a fast, high precision and good match-adaptive image registration method.In the article, it implements image registration between the human torso X-ray transmission image and the3D CT data base on mutual information measure. In the experiment of image registration, the Powell algorithm and Ant colony algorithm are applied as the optimal search strategy to obtain the corresponding transformation parameters. Powell algorithm is a classic local optimization algorithm. Due to the limitations of the algorithm, there is a certain probability that it cannot obtain the local optimum value when the objective function is not ideal and contains lots of local extrema such as human skeleton medical images. In this paper, we propose a2D-3D image registration method base mutual information and ant colony algorithm. Through several comparative experiments show that the improved algorithm can achieve medical image registration better. It is not only to solve the problem that the registration parameters are not the global optimum, but also achieve more accurate registration. On the other hand, in order to solve the problem that the intelligent algorithm needs to do more calculations and spend more time, we introduce a multi-resolution processing method that solving the problem from rough to fine. Combined with the principle of wavelet transform technology, we decompose the images twice and the registration parameters can automatically adjusted in different levels. The result shows that this method can effectively reduce the registration time and improve the efficiency of registration.We have done some useful exploration in researching the method of registration between2D and3D medical images, and achieved some results. This result has a certain practical value to improve the efficiency of surgical navigation.