Multi-modal Medical Image Registration And Fusion Algorithms

With the development of medical image engineering and computer technology, medical imaging has become one of the most important parts in modern medical treatment. There are medical images of multi-modality in clinical diagnoses which obtained by different imaging principles, such as Computerized Tomography (CT), Magnetic Resonance Imaging (MRI), Single Photon Emission Computed Tomography (SPECT) etc. These images contain different information about human viscera and pathological tissue. In the area of image processing, image fusion is to merge medical images of different modality into one image in image processing. The fused image can integrate all kinds of image information, precisely and roundly show the tissue and apparatus structure, function and the pathological changes, and integrate anatomical information and functional information. It is convenient for doctors to obtain synthesis information of pathological tissue and apparatus, and make more exact diagnosis and choose the best treatment.We need to solve the problem of image registration before the quantitative analysis of different images. The task of medical image registration is to find a geometric relation or transformation among the same regions of interest in two or more images. The effect of image registration directly affects the qualities of fusion image, so both the exact registration and proper fusion are necessary for doctors to obtain the exact diagnosis results.Based on the theories of morphology gradient, normal mutual information, wavelet packet and self-adaptive operator, this thesis makes a deeper research on multi-modality medical image registration and fusion. The main contents of this thesis are taken as follows:Firstly, medical image registration is studied in this thesis. In the research, an improved method of medical image registration based on morphology gradient and mutual information is provided. The new method can shorten the running time and amend the traditional local maximum problem by using intensity information and space geometric figure. CT, MRI and SPECT-three modalities medical images are simulated in MATLAB. The results of experiment show that the accuracy and running speed of new method is superior to that of the traditional mutual information method, and the effect of noise is obviously weakened.Secondly, this thesis gives a detailed research about multi-modality medical image fusion on the basis of image registration. In the research of medical image fusion, this thesis provides a self-adaptive medical image fusion algorithm based on wavelet packet. Firstly decompose medical images by wavelet packet, and then process the sub-images through the adaptive operators, finally get the fusion image by wavelet packet reconstruction. Wavelet packet can decompose the images frequency band in different level, and can decompose high frequency band in a higher scale which the wavelet decomposition couldn't. So the wavelet packet can improve the frequency resolution. Self-adaptive operator can adjust the powder weight coefficient automatically according to the images information, so it can avoid the error brought by setting fastness powder weight coefficient. Three fusion methods which are Laplacian Pyramid transform, wavelet transform and wavelet packet transform are simulated in MATLAB. We choose average gradient, mean, standard deviation, information entropy and correlation coefficient as criterions to evaluate the qualities of fused images in the thesis. We can get the results by comparing the evaluate parameters, and the results of experiment show that the self-adaptive medical image fusion algorithm based on wavelet packet is obviously better than the two others.Finally, we make a conclusion and propose the future research directions in this field.