The Technology Of Medical Image Fusion

Medical image fusion, a subject which involves information science and medical imaging science, is a new hot spot in medical image processing. As the basis of medical image fusion, the preprocessing and registration of images have very important meaning. The effect of preprocessing and the precesion of registration will influence the effect of images fusion directly.Some basic 2D medical image processing methods related with image fusion such as image interpolation, spatial transformation, image enhancement are compared in the paper firstly.Secondly, the paper focuses on the multimodality medical image registration based on mutual information. It's a full-automatic method that uses mutual information as the matching criterion of corresponding pixels in both images. The mutual information is assumed to be maximal if the two images are geometrically aligned. The optimal rotation and translation parameters are searched by simplex algorithm. The experiment result shows this method has good fusion effect.Thirdly, the image fusion methods based on space domain, pyramid decomposition and wavelet transform are studied. The methods based on space domain operate pixels directly and are easy but week in fusion results. The methods based on pyramid decomposition are multiscale and multiresolution. Some of these methods are studied and the fusion effect of them are compared. The wavelet transform is a method of multiscale and multiresolution, it is used in signal analysis and image processing. We study the methods of image fusion based on wavelet transform, introduce some fusion rules based on pixels and windows, and present a new fusion rule. According to evaluation criteria that are given in the paper and a series of experiments results, what can be done to fusion performance by wavelet radix, the number of decomposition levels, different fusion rules and the size of windows for measure are studied and some useful conclusion are got. The experiment result also shows that the improved fusion rule can preserve details of source images effectively at the same time.At last, a medical images preprocessing and fusion software is developed on the platform of VC++.