Research On Medical Image Fusion Methods Based On Curvelet Transform

In modern society, with the improvement of people’s living standard, citizens have paid more and more attention to health care, which not only promotes the rapid development of modern medical science, but also leads to the wide application of medical imaging methods. The detail information of human tissues and organs can be expressed in multi-source images, which is an important basis for clinical diagnosis and treatment. Usually in the diagnostic process, in order to accurately determine the focus location and the stage condition of disease, information contained in multiple medical images must be integrated and comprehensively considered. Medical image fusion technology can transfer multiple medical images into a complete image with certain algorithm. And it is definitely the inevitable requirement of the integrated medical diagnosis. Medical image registration is the precondition of fusion. It provides pre-processing to multi-modality medical image through translating and rotating operation. To a certain extent, registration can avoid the mutual interference of different organ information in the process of fusion.In the medical image registration part of this paper, from the basic starting point of clinical application, the algorithm achieves a two-dimensional rigid image registration, analyzes the medical image registration method based on mutual information, accomplishes a comprehensive study of the effect of interpolation and optimization algorithm of mutual information casting on processing results. Aiming at accelerating the matching speed, this paper takes the method of custom-spindle initial registration of medical images. Aiming at the problem of local minimum or maximum in mutual information registration method, this paper puts forward an improved partial volume interpolation method, and by introducing the concept of gradient mutual information. So that the accuracy of image registration has achieved a new level. Finally, through the comparison of simulation, it is can proved that the brand new GSI-CSPV-PSO combining algorithm, has great advantages in medical image registration.Aiming at the registrated medical images, this paper describes a digital implementation of a new mathematical transform—the second generation Curvelettransform, and discusses both the principles and the ways of calculation. Based on this algorithm, the paper analyzes the Curvelet coefficients through the simulation completed in this paper, and successfully performs the reconstruction of image. This paper introduces Curvelet transform into practical applications: firstly, using Curvelet transform to decompose the source images; secondly, combining the coefficients of different scales with specialized fusion rules. Thus, the proposed method can get fusion result through the process of image reconstruction. Finally, through contrast experiments, this paper proves that modified Curvelet transform in medical image fusion has good practicality in improving the fusion result.