多通道脑电信号中眼电伪迹的自动识别与去除方法研究

Automatic identification and removal of oculoelectric artifacts in multichannel EEG signals

目的:针对当前眼电伪迹去除算法会带走有效脑电信号的情况,提出一种快速独立成分分析(fast indepen-dent component analysis,FastICA)与经验小波变换(empirical wavelet transform,EWT)算法相结合的眼电伪迹识别与去除方法。方法:首先通过FastICA和自适应样本熵筛选出含有眼电伪迹的独立成分分量;其次通过EWT算法和自相关系数剔除眼电伪迹成分,保留有用的脑电信号;最后进行EWT逆变换,合成成分分量,并与不含眼电伪迹的独立成分分量进行重构,得到眼电伪迹去除后的脑电信号。通过自采集数据集与公开数据集验证眼电伪迹的去除效果。结果:该方法能够自动识别与去除眼电伪迹并保留有效脑电信号,且能针对不同被试者个体的差异性进行自适应。结论:该方法具有鲁棒性强、准确率高的优点,能够较好地识别并去除多通道脑电信号中的眼电伪迹。

Objective To propose an oculoelectric artifact removal method by combining fast independent component analysis(Fast ICA)and empirical wavelet transform(EWT)algorithms to solve the problem of the existing algorithms in mis extracting valid EEG signals.Methods Firstly,the independent components containing oculoelectric artifacts were filtered by FastICA and adaptive sample entropy.Then,the EEG artifacts were removed by empirical wavelet transform and autocorrelation coefficient,and the useful EEG signals were retained.Finally,the inverse empirical wavelet transform and the independent components without oculoelectric artifacts were used to reconstruct the EEG signal after the removal of oculogram artifacts.The validation of the oculoelectric artifact removal effect was performed by experiments on the self-test dataset and the public dataset.Results The method proposed could automatically identify and remove the oculoelectric artifacts and retain the effective EEG signals,and adapted to the individual variability of different subjects.Conclusion The method has the advantages of high robustness and accuracy,and can intelligently and accurately identify and remove the oculoelectric artifacts from multi-channel EEG signals.