The Research Of Multimodality Medical Image Registration And Wavelet Transform-based Image Fusion Technology

With the development of the science theory and manufacture, more and more medical imaging equipments are equipped in hospitals. Each kind of imaging technology has the different specialty and sensitivity to different tissues. Multi-modality medical image fusion technology is defined as using image fuse technology in medical images, in order to get new fusion image to supply more abundant pathology information than both of two sources of images for clinic diagnoses. Before image fusion, lots of processes should be done firstly,the image registration is the most important process.Image registration is one of the most important processes in the image processing. It has an important place in image identification,target detection, analysis of sequential images, image fusion, etc. Now there are lots of registration algorithms for multi-modality image. Usually the common image registration algorithms are classified into three sorts: intensity-based image registration algorithm, feature-based image registration algorithm and decision-based image registration algorithm. The merit of intensity-based image registration algorithm is high registration ratio, but because most of the image pixels will be used in the processing of registration, this algorithm is in low velocity and sensitive to noise. So far a lot of techniques are introduced to ameliorate the intensity-based algorithm,in this paper, we present an improved algorithm based on pyramid algorithm; feature-based image registration algorithm uses the features selected from source images rather than most of the pixels to finish the process, so the velocity of computation is very quick, though high complexity of feature selection algorithm and low registration ratio are its disadvantages.According to the specialties of the first and second algorithm, the more popular way is synthesizing both of the two algorithms, executing intensity-based image registration algorithm at first to get the rough result, then do image registration process by feature-based image registration algorithm. Secondly under the guide of the last result. the third method can achieve the balance between velocity and registration ratio. According to the theory, we present an algorithm which is based on the image moments and multi-information theories to fix the desires of medical image registration.This paper introduces the principles of image decomposition and reconstruction based on wavelet-pyramid and image fusion algorithm; we mainly discuss the fusion principles of weighted fusion operators and region-based algorithm. We have done the CT and MRI skull images fusion, and got ideal result, measured the effect of medical image registration.We developed an application software called "Multi-modality medical image registration and fusion system" based on VC++6.0, for the medical image registration and fusion algorithms in this paper.