Abstract:

In order to improve the problems existing in the traditional Independent Component Analysis（ICA） algorithm for automatic removal of Ocular Artifact（OA）, such as slow speed of recognition of OA, acquisition of synchronous reference EOG signals, and loss of EEG signals, a recognition method without reference Electro-Oculogram（EOG） signalsis proposed, which can automatically removal OA. Firstly, using the FastICA into independent component, calculate the spectrum energy entropy of each independent components, with energy spectrum entropy as the criterion to identify the OA components. Then the EEG signals in the OA component are isolated by the peak window and spliced with other independent components. Further more, the clean EEG signals are restored with the inverse algorithm of FastICA. These experimental results show that this method can quickly, precisely and automatically remove the OA and retain other EEG components. The average time of spectral energy entropy for the recognition of OA is 0.01 s, and the accuracy rate is 98%, which is suitable for real-time EOG removal.

Key words: Electroencephalography（EEG）, Ocular Artifact（OA）, energy entropy, peak window

摘要：

为改进传统独立分量分析自动去除眼电伪迹算法中存在识别眼电分量速度慢、需采集同步参考眼电信号、丢失脑电信号问题，提出一种不需要参考眼电信号的眼电伪迹自动识别去除方法。利用FastICA分解出独立分量，计算各独立分量频谱能量熵，以频谱能量熵值作为判据识别出眼电分量；然后使用峰值窗口分离出眼电分量中存在的脑电信号，与其他独立分量进行拼接；利用FastICA逆变换重构出去眼电伪迹的脑电信号。实验结果表明：该方法能准确快速自动地去除眼电伪迹，并较好地保留其他的脑电信号成分；频谱能量熵识别眼电伪迹平均用时为0.01?s，准确率为98％，适用于实时EOG去除。

关键词: 脑电信号（EEG）, 眼电伪迹, 能量熵, 峰值窗口