The Application Of Independent Component Analysis Algorithm In Functional Magnetic Resonance Imaging Data

The human brain is very complicated, and people have been trying to understand it forso many years. The development of brain imaging technology has made people, based on theinitial anatomical localization, conduct further tests on the basic process of brain activities.The Functional Magnetic Resonance Imaging (the FMRI) is a new technology based onblood oxygen level, which depends on imaging and blood flow sensitivity. The FMRI showsmany advantages such as having no radiation or traumatic to human body, presenting highresolution and being easy to be conducted repeatedly regardless of time and space. Duringrecent years, it has been chosen as the best way to research the brain activities. Consequently,people have invented many processing methods based on the FMRI. This article focuses onthe application of the FMRI analyzing a new and independent component method.The previous researches on the processions of FMRI data were conducted, inaccordance to the order of time and based on the pre-understood situation, so as to get brainactive region. The stimulation of the brain and its influential factors are unknown without thecollection order activities, which means that the signal source and mixed matrix areunknown. That is typical blind source separation. To solve the problem, we should choosethe most commonly used method of independent component analysis (the ICA). By applyingthe ICA, we can realize the hybrid separation of the data and get components related toextraction task.This paper first introduces the application of some estimation algorithm methods basedon the independent component analysis to process FMRI data. Through the introduction ofthe common FASTICA algorithm, we introduced a new algorithm—the ROBUSTICAalgorithm based on kurtosis. With certain objectives, we designed a series of experiments,and then we analyzed and processed the FMRI data. With the help of SPM software, weapplied the ICA (including ROBUSTICA algorithm and FASTICA algorithm) as well as GIFT software to separate the FMRI data components. By comparing the data with those gotthrough FASTICA algorithm, we found that the new algorithm is more advanced. It showedsome advantages. Firstly, the algorithm can improve robustness when bad points appearamong signals. Secondly, the algorithm has, to some extent, improved calculation speed.1)the algorithm can improve.2) this algorithm in a certain extent appropriate increase the, withthe new algorithm combining composition analysis method to group of FMRI data analysis.