一种改进FastICA算法及其在fMRI数据中仿真应用.docVIP

一种改进的FastICA算法及其在fMRI数据中的仿真应用 　　摘 要：为了更有效地去除功能磁共振成像（functional magnetic resonance imaging，fMRI）数据中所包含的大量噪声信号，研究了快速独立分量分析（Fast Independent Component Analysis，FastICA）算法，针对该算法在处理低信噪比的fMRI数据时估计精度降低问题，采用修正白化矩阵方法改进FastICA算法，与传统的FastICA算法分别对fMRI数据进行去噪仿真分析。对比两种方法去噪结果，发现改进的FastICA算法能更有效地去除fMRI数据中的噪声信号，为剔除fMRI伪迹，提高fMRI数据预处理效果奠定了坚实的科学基础。 　　关键词：功能磁共振成像；FastICA；白化矩阵；仿真 　　DOI：10.11907/rjdk.172553 　　中图分类号：TP312 　　文献标识码：A 文章编号：1672-7800（2018）004-0067-04 　　Abstract：In order to more effectively remove the large amount of noise signals contained in the functional magnetic resonance imaging（fMRI） data，Based on Fast Independent Component Analysis （FastICA） algorithm research in this paper， according to the algorithm when dealing with fMRI data of low signal-to-noise ratio estimation precision degradation，correction of albino matrix method is used to improve the FastICA algorithm，at the same time the paper with the traditional FastICA algorithm of denoising simulation Analysis of fMRI data respectively. By comparing the results of the two methods，it is found that the improved FastICA algorithm can effectively remove the noise signal in fMRI data， which proves the effectiveness of the proposed method.This lays a solid scientific foundation for eliminating fMRI artifacts and improving the preprocessing effect of fMRI data. 　　In order to have more effective removal of large amount of noise signals in functional magnetic imaging （FMI）， we research on Fast Independent Component Analysis （Fast ICA） algorithm. Aiming at estmation precision degradation of the algorithm when dealing with FMRI data， we employ the improved whitening matrix to improve FastICA algorithm. Denosing simulation analysis is made by the improved and the traditional FastICA algorithm respectively. By comparing the results of the mentioned analysis we find the improved algorithm can improve the improved FastICA algorithm can effectively remove the noise signal in FMRI data to eliminate FMMRI artifacts， which forms a