脑-机接口中空域滤波技术现状与进展

作为一种特殊的人-机交互模式,脑-机接口(brain-computer interface,简称BCI)技术已成为当前脑科学和智能信息处理领域的研究热点.其中,基于头皮脑电(electroencephalography,简称EEG)的BCI(EEGBCI)技术因具有良好的安全性和可操作性,吸引了研究者的广泛关注.但头皮EEG非常有限的空间分辨率和易受干扰等特性,很大程度上限制了EEG-BCI技术的实用化进程.因此,EEG信号处理和模式识别新方法研究已成为BCI领域的一个关键问题.在现有的信号处理方法中,空域滤波技术在EEG伪迹消除和任务相关神经活动获取方面表现出了较明显的优势,近年来在EEG-BCI系统实现研究中得到了广泛应用.论文以运动想象BCI(motor imagery BCI,简称MIBCI)为应用背景,对独立分量分析(independent component analysis,简称ICA)和共同空间模式(common spatial pattern,简称CSP)两种代表性空域滤波方法的原理及其性能进行介绍、分析和比较,总结出两种方法各自的优势和不足,并给出了改进思路.同时指出,ICA空域滤波方法在运动想象脑-机接口系统实现中更具应用潜力.

Recent progress and future development of spatial filtering technique in brain-computer interface

As a special mode of human-machine interface,brain-computer interface (BCI) has gradually become a hot research topic in brain science and intelligent information processing,among which the scalp electroencephalography (EEG)-based BCI has attracted wide attention of researchers due to its good security and operability.Whereas,some EEG features,such as limited spatial resolution and high susceptibility to interferences,block the practical process of EEG-BCI technology.Therefore,new EEG signal processing technology and pattern recognition method have been always key issues in BCI field.In state-of-the-art signal processing methods,spatial filtering methods have obvious superiority in artifacts elimination and task-related neural activities acquisitions,so as to have bccn used more and more widely in EEG-BCI system implementation.This paper took motor imagery BCI(MIBCI) as application background,and principles of two representative spatial filtering methods of independent component analysis (ICA) and common spatial pattern (CSP) were introduced,performances of which were analyzed and compared.We summarized advantages and disadvantages of two methods,and the corresponding improvement solutions were proposed.Meanwhile,it could be concluded that ICA spatial filtering method possessed the better application potential in MIBCI system implementation.