基于独立向量分析的脑电信号中肌电伪迹的去除方法

脑电数据经常被各种电生理信号伪迹所污染。在常见伪迹中，肌电伪迹特别难以去除。文献中最常用的方法包括诸如独立分量分析(Independent Component Analysis, ICA)和典型相关分析(Canonical Correlation Analysis, CCA)等盲源分离技术。该文首次提出一种基于独立向量分析(Independent Vector Analysis, IVA)的新方法，用以去除脑电中的肌电伪迹。IVA同时使用高阶统计量和二阶统计量，因此该方法能够充分利用肌电伪迹的非高斯性和弱相关性，兼具ICA方法和CCA方法的优势。实验表明，使用IVA方法可以在保留脑电成份的同时极大抑制肌电伪迹，效果显著优于ICA法和CCA法。

Removal of Muscle Artifact from EEG Data Based on Independent Vector Analysis

ElectroEncephaloGram (EEG) data are often contaminated by various electrophysiological artifacts. Among all these artifacts, removing the ones related to muscle activity is particularly challenging. In past studies, Independent Component Analysis (ICA) and Canonical Correlation Analysis (CCA), as Blind Source Separation (BSS) methods, are widely used. In this work, a new method for muscle artifact removal in EEG data using Independent Vector Analysis (IVA) is proposed. IVA utilizes both the higher-order and second-order statistics, so that it makes full use of non-Gaussianity and weak autocorrelation of the muscle artifact and has the advantages of both ICA and CCA. The proposed method is examined on a number of simulated data sets and is shown to have better performance than ICA and CCA. The proposed IVA method is able to largely suppress muscle activity and meanwhile well preserve the underlying EEG activity.