多通道脑电信号眼电伪迹自适应去除方法

针对现有方法在眼电伪迹自动去除中存在有用信息丢失，伪迹分量识别困难的问题，提出了一种结合粒子群优化算法、独立成分分析和小波变换的伪迹自适应去除算法。首先，采用均方根误差和Pearson相关系数设计了粒子群优化算法的适应度函数，利用优化算法实现了两个样本熵阈值的自适应设置；然后利用快速独立成分分析算法将脑电信号分解为统计独立分量，根据第一个样本熵阈值自动识别含伪迹分量，含伪迹分量经过四层小波分解得到五个小波分量，根据第二个样本熵阈值自动识别伪迹分量，将识别的伪迹分量置零；最后经过小波重构和逆变换，获得去除眼电伪迹的脑电信号。采用Graz data set A数据集进行实验验证，结果表明提出的方法能够实现多通道脑电信号伪迹的自动去除；采用Klados数据集进行实验验证，结果表明，与SE-CEEMDAN方法相比，采用提出方法实验获得的均方根误差降低了4.816，约38.2%，Pearson相关系数提高了0.025，约2.97%。

Automatic removal method of EOG artifacts for multi-channel EEG

Aiming at the problem of the existing methods in the automatic removal of EOG artifacts that useful information is lost and the recognition of artifact components is difficult, an automatic artifact removal algorithm combining particle swarm optimization algorithm, independent component analysis and wavelet transform is proposed. First, the fitness function of the particle swarm optimization algorithm is designed by using the root mean square error and Pearson correlation coefficient, and the adaptive setting of the two sample entropy thresholds is realized by the optimization algorithm. Then the fast independent component analysis algorithm is used to decompose the EEG signal into statistical independent components, and the artifact components are automatically identified according to the first sample entropy threshold. The artifact components are decomposed by four layers of wavelet to obtain five wavelet components. The artifact components are automatically identified according to the second sample entropy threshold, and the identified artifact components are set to zero. Finally, after wavelet reconstruction and inverse transformation, the EEG signal with EOG artifacts removed is obtained. The Graz data set A is used for experimental verification, and the results show that the proposed method can realize the automatic removal of EEG artifacts from multi-channel EEG signals. The Klados data set is used for experimental verification. The results show that, compared with the SE-CEEMDAN method, the root mean square error obtained by the proposed method is reduced by 4.816, about 38.2%, and the Pearson correlation coefficient is increased by 0.025, about 2.97%.