一种针对EEG伪迹的多窗卷积压制算法

A Multi-Window Convolution Suppression Algorithm for EEG Artifacts

脑电信号(EEG)会受到伪迹的严重干扰,可以利用独立成分分析分离出观察信号的源成分,并根据源成分的时、空等特征区分神经信号成分和伪迹成分,随之删除伪迹成分可恢复较干净的观察信号。但在删除整个伪迹成分的同时势必会丢失一些有用的神经信号,这会造成神经信号泄露。为此,提出一种降低伪迹噪声和减少神经信号泄露的均衡方法,在盲源信号分离的基础上,利用局部化的时间特征二分类伪迹成分和神经信号成分,以多种窗函数组合优化并卷积压制局部伪迹成分。通过对观察信号计算信号伪迹比(SAR),8导、32导非校正脑电信号SAR分别为-19.765dB、-19.016dB,校正后8导、32导脑电信号SAR分别提高到2.832dB、2.743dB。结果表明该方法可以消除眼电等伪迹且降低神经信号泄露的影响。

Electroencephalogram(EEG)can be seriously interfered by the artifact,so the source components of the observed signal can be separated by independent component analysis,and the neural signal components and artifact components can be distinguished according to the time and space characteristics of the source components.However,when deleting the whole artifact,some useful neural signals will be lost,which will cause the leakage of neural signals.Therefore,it is necessary to find a balanced method to reduce artifact noise and nerve signal leakage.Based on blind source signal separation,the artifact components and neural signal components are locally classified by time fea ture dichotomy,and multiple window functions are combined and convolved to suppress the local artifact components.SAR was calculated based on the observed signal.The 8-guide and 32-guide non-corrected EEG signal SAR were-18.7268dB and-19.016dB,respectively.After correction,the 8-guide and 32-guide EEG signal SAR were improved to 1.9414dB and 1.9042dB.The results show that the proposed method can eliminate visual artifacts and reduce the influence of nerve signal leakage.