Image Registration Algorithm And Its Application In FMRI Preprocessing

Recently with the rapid development of medical imaging technology,a functional magnetic resonance imaging(fMRI)technology develops on the basis of the magnetic resonance imaging.Due to its non-invasive,accurate location of brain function,and good repeatability and feasibility,it has been widely concerned and studied.However,considering the influence of imaging equipment,individual differences and head movement in this type of study,the collected fMRI images can not be directly processed and analyzed.A series of preprocessing operations must be performed first to eliminate the effects of these factors.Among them,image registration plays an important role,and its results are directly related to the accuracy of the final location of brain function.In addition,when modeling the fMRI data,there is often a certain collinearity between the regression of the model,which leads to inaccurate or even erroneous statistical results.Therefore,the correct processing and analysis of fMRI data is the basis and prerequisite for accurate localization of brain function.In this paper,based on the fMRI image as the research object,a further and comprehensive research about the problems existing in the process of fMRI data processing and analysis are discussed and studied.By revealing the underlying mechanism of brain cognitive activity,it is of great significance to the diagnosis and treatment of many major brain diseases,as well as related pathology and pharmacology stdudies.The main research work of this paper are as follows:Firstly,the paper introduces the basic concept and framework of the image registration algorithm,expatiates the registration process from the four components: geometric transformation,feature space,similarity measurement and search strategy.And we take the actual fMRI data as an example,making a detailed elaboration of the general process and statistical analysis process by using SPM software package.Then,considering the influence of the head movement in the process of fMRI data acquisition,a functional magnetic resonance image registration method based on mutual information is proposed.In order to further improve the speed and accuracy of registration,the spindle method and multi-resolution strategy are also used.Finally,Concerning the collinearity between the regressors in fMRI-based general linear model,a method of orthogonalization is proposed.It can decompose the collinear regressors into independent part of each other,so as to eliminate the effect of one regression on the other regression in the result variable,which can obtain more accurate results.