基于CEEMDAN-ICA的单通道脑电信号眼电伪迹滤除方法

传统盲源分离法不能解决欠定问题,且分离信号与源信号对应关系不确定.提出一种基于自适应噪声完备经验模态分解(CEEMDAN)和独立成分分析(ICA)相结合的脑电信号眼电伪迹自动去除方法.该方法首先将含伪迹脑电信号自适应分解成多维本征模态函数(IMF),以满足盲源分离方法对信号正定或超定要求,再对本征模态函数用ICA方法构建多维源信号,最后利用模糊熵阈值判据判别多维源信号中的伪迹信号,完成滤波并重构脑电信号.该方法相比于其他算法,能更好的去除眼电伪迹并保留原始信息,适合单通道脑电信号预处理.

Electroencephalogram artifact filtering method of single channel EEG based on CEEMDAN-ICA

The traditional blind source separation method can not solve the underdetermined problem, and the corresponding relationship between the separated signal and the source signal is uncertain. An automatic removal method of EEG artifacts based on the combination of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and independent component analysis (ICA) is proposed. This method firstly decomposes EEG signals containing artifacts into multi-dimensional Intrinsic mode functions adaptively to meet the requirements of positive or overdetermined signals in blind source separation model, and then builds multi-dimensional source signals by using ICA method for intrinsic mode functions. Finally, we use the threshold criterion based on fuzzy entropy to filter the artifact signals in the multi-dimensional source signal and reconstruct the EEG signal. Compared with other algorithms, this method can better remove the eye artifacts and retain the original information, which is suitable for single-channel EEG signal preprocessing.