摘要
独立成份分析(ICA)是信号处理领域中新近发展起来的一种很有应用前景的方法,而脑功能磁共振(fMRI)信号的有效分离与识别是一个正在研究和试验之中的技术领域。因此,发展基于ICA的fMRI数据处理方法具有明显的理论价值和应用前景。本文首先介绍了ICA原理,分析了现行ICA-fMRI方法采用的信号与噪声的空域分布相互独立的信号模型所存在的明显不足,然后提出了微域中的信号与噪声的时域过程相互独立的fMRI信号模型熏从而建立了一种新的fMRI数据处理方法:邻域独立成份相关法。合理的fMRI实验数据处理结果验证了新方法的合理性。
Independent component analysis (ICA) is a newly developed promising technique in signal processing applications. The effective separation and discrimination of functional magnetic resonance i-maging (fMRI) signals is an area of active research and widespread interest.Therefore-the development of an ICA based fMRI data processing method is of obvious value both theoretically and in potential applications. In this paper- introduced ICA principles and analyzed firstly is the drawback of the extant popular ICA-fMRI method where the adopted signal model assumes the independence of spatial distributions of the signals and noise. Then presented is a new fMRI signal model- which assumes the independence of temporal courses of signal and noise in a tiny spatial domain. Consequently we get a novel fMRI data processing method: Neighborhood independent component correlation algorithm. The effectiveness is elucidated through a real fMRI data test is presented.
出处
《中国医学物理学杂志》
CSCD
2003年第4期245-248,共4页
Chinese Journal of Medical Physics
基金
国家自然科学基金项目(30200059)
重点项目熏教育部科学技术研究重点项目(02065)
四川省杰出青年学科带头人培养基金项目
电子科技大学青年科技基金项目yf021103
关键词
脑功能磁共振成像
独立成份分析
空域分布
时域过程
信号模型
functional magnetic resonance imaging (fMRI)
independent component analysis (ICA)
spatial distribution
temporal process
signal model