Dynamic Brain MR Image Registration Algorithm Based On Evolutionary Computation

Neurodegenerative disease has become one of the main threat of the health. Theearly diagnosis can greatly avoid the incidence of disease. MRI method is one of theeffective treatments for early diagnosis of neurodegenerative disease. The registrationmethod is the key step, which directly influences clinical application of MRI processing.So the research on registration method is a key problem for improving early diagnosisof neurodegenerative disease based on MRI.The research is under the support of Central College Fund.The major contributionof the thesis is as follows:Conducted research on the basic theory of linear decrease weight particle swarmoptimization algorithm and proposed a new dynamic brain MR image regsitrationmethod. The method has two forms which is full inheritance and half inheritance, wecall them LDWPSO-FI-DIR and LDWPSO-HI-DIR algorithms respectively. The twoforms optimize initial population, overcame the flaw of too long time when registeringmany images continuously. Make discussion and analysis on the results of theregistration algorithms.Conducted similar research on the genetic algorithm (GA) and proposed a similardynamic brain MR image registraiton method based on GA. The method has two formswhich is full inheritance and half inheritance, we call them GA-FI-DIR and GA-HI-DIRalgorithms respectively. Make discussion and analysis on the results of the registrationalgorithms.compared the performances of the algorithm under different mutual informationmeature. Make discussion and analysis about the difference between the two dynamicoptimization algorithms under difference similarity measures based on mutualinformation.The structure of this paper is as follows: chapter one introduces the background,significance and the domestic and foreign current research situation, and it alsointroduces the main research contents and related work; chapter two introduces brainframework and theoretical basis of MR image registration method, including basicconcept, key technology; chapter three do research on the basic theory of linear decreaseweight particle swarm optimization algorithm. And the new dynamic brain imageregistration algorithms are presented. This chapter includes theory analysis, the algorithm and the relevant experiments; chapter four conducted researchs on the basictheory of genetic algorithm and a new dynamic brain MR image registration algorithmbased on GA is proposed. This chapter includes theory analysis, the algorithm and theanalysis of experimental results; chapter five compares registration results of differencedynamic algorithms based on difference mutual information measures; chapter sixillustrates the major contribution of the paper and the future work.