中文摘要：

目的：探讨功能磁共振信号的主要组成及运动伪影对其影响。方法：采用自编程序实现时间聚类分析算法对2例事件相关模式视觉刺激任务的功能磁共振数据进行分析，并与（statistical parametric mapping，SPM）拟合曲线图及实验时间序列中运动轨迹记录做比较分析。结果：时间聚类分析在血氧水平反应阈值范围内可检测到大量反应信号，通过与SPM拟合调整曲线图比较发现，BOLD事件反应时，与时间聚类分析（TCA）结果可发现与其重叠的信号；通过与时间序列中运动轨迹记录进行比较发现，TCA所检出较大的信号均与运动波峰有很大程度的相关。结论：时间聚类分析作为一种数据驱动的功能磁共振数据分析方法，一定程度上可以反映真正的血氧水平，但受运动伪影影响严重，在类短时程刺激的事件相关模式的功能成像实验时可谨慎使用。

英文摘要：

Objective：To investigate the components of the signal resulting from the data processed by temporal clustering analysis （TCA） and the affection of motion artifact. Methods：The algorithm of temporal cluster analysis implemented by home-made software was used to process two sets of ftmctional MRI data of event-related paradigm when visual stimulating. And some comparisons were made in fitted responses, TCA motion trochoid and results. Results：Lots of signal was detected by TCA with threshold that during the random of hemodynamic response. When comparison was made between TCA and SPM fitted responses, the signal of real BOLD response could be found in TCA results. And when between TCA and motion trochoids, There was highly correlation between them. Conclusion：TCA as a method of data driven model, was able to detect some real BOLD signal, but the motion artifact was rather awful. Which suggests it must be cautious to use TCA when performing the fMRI experiment resembling event-related paradigm with short stimulus.