独立成分分析的fMRI对癫痫灶检测的对比研究

Comparative study of Independent Components Analysis and Combined EEG Functional MRI in Epileptic Focus Detection

目的采用独立成分分析(independent component analysis,ICA)方法,对局灶性癫痫的功能磁共振成像(functional magnetic resonance imaging,fMRI)数据进行分析,评价其在fMRI癫痫灶定位研究中的应用价值。方法采用经预处理的空间ICA方法,对12例局灶性癫痫患者18组数据进行处理。利用脑电同步功能磁共振成像(sim-ultaneous electroencephalogram and functional MRI,EEG-fMRI)检出的间期痫样发放假设驱动模式设计,对检出的各独立成分的时间序列进行多元回归排序,观察癫痫数据ICA结果的空间及时间特征。并与传统广义线性模型(general-linear model,GLM)方法进行对比,评价ICA对癫痫静态fMRI检出的效能。结果与传统GLM脑电同步功能成像方法对比(10/18),ICA方法具有较好的癫痫活动检出能力[传统广义线性法:55%(10/18);ICA法72%(13/18)]。此外,ICA方法还可以发现缺省网络等静息态脑认知网络受癫痫活动发放影响的情况。结论 ICA具有检出静态fMRI数据中间期痫样放电(interictal epiletiform discharges,IEDs)信号的能力,在癫痫活动的研究方面具有良好的应用前景。

Objective To research the functional MRI in focal epilepsy using the method of independent components analysis,and to estimate its application value in the research of detecting epileptic focus.Methods The resting functional MRI data of 18 sessions of 12 focal epileptic patients were analyzed by spatial independent components analysis after preprocessing.The components of ICA were sorted according to the time course multiple regressions with hypothesis-driven mode matrix,which was designed by the interictal epileptiform discharges（IEDs） points detected with simultaneous EEG functional MRI.The spatial and temporal character of the components of interesting were investigated.Results Compared with conventional general-linear model（GLM） approach,ICA was found to have better capability to detect epileptic IED-related BOLD activation（GLM vs ICA： 55% vs 72%）.Moreover,ICA could detect abnormalities in the resting-state default-mode network in epilepsy.Conclusion The method of independent analysis has the ability to detect the BOLD activation of IEDs of focus epileptic patients,which has a good perspective in the application of the epilepsy research.