Research Of Medical Image Registration Algorithm

Medical image registration is an important technology in clinical medical analysis,an important step to complete computer-aided diagnosis,and also a prerequisite for the important technology of three-dimensional reconstruction of medical images.Therefore,medical image registration is also a hot research issue in medical image processing.Medical image registration is to find out the linear or nonlinear transformation between two medical images in space and mode by means of operation,and to seek the mapping relationship between them,so as to achieve the purpose of information fusion.With the rapid development of computer technology and assisted medical diagnosis technology,the study of medical image registration has been paid more and more attention by researchers in the corresponding field.There are many different kinds of image registration,according to the classification of the above methods can be divided into eight kinds roughly,but the registration process is roughly same,its steps roughly consists of four parts,preprocessing part: pretreatment is to make a noise of image processing,first remove the noise in the image,and then use the normalized regularization method to deal with image pixels,the alignment algorithm based on image.Image feature space measurement and matching: Find the mapping relationship between images on pixels or other features.Construction and optimization of energy function: The energy function which is constructed according to the information to be registered is obtained according to the optimization method.Discussing the main process of the registration algorithm used by the article,throwing the rigid on time use for optimization model,according to the process,start with coarse segmentation is rigid registration,first of all,to explore a kind of took a multiresolution fusion the mutual information function and difference of the sum of the squares of the rigid registration,then the large deformation is derived from the perspective of mathematical differential homeomorphism registration algorithm framework,In this paper,a registration method based on large deformation differential homeomorphism is discussed,and the registration of two and three dimensions is realized.In the end,the method of multi-atlas fusion is explored for image segmentation,and the image after registration is fused with certain rules to obtain more accurate segmentation results,and the automatic operation of this kind of method is realized.