ICA去除EEG中眼动伪差和工频干扰方法研究
he Method Research of Applying Independent Component Analysis
眼动伪差和工频干扰是临床脑电图(EEG)中常见噪声，严重影响其有用信息提取。本文尝试采用独立分量分析(Independent Component Analysis, ICA)方法分离EEG中此类噪声。 通过对早老性痴呆症(Alzheimer disease, AD)患者临床EEG信号（含眼动伪差和混入工频干扰，信噪比仅0dB）作ICA分析，比较了最大熵(Infomax)、扩展最大熵(Extended Infomax) 等ICA算法和奇异值分解(singular value decomposition, SVD)方法的分离效果，证实虽然最大熵算法可以分离出眼动慢波，但难以消除工频干扰，为此需采用扩展的最大熵算法；并知ICA方法在极低信噪比时也有较好的抗干扰性，且在处理非平稳信号时有好的鲁棒性；而常用的SVD技术则略逊一筹；文中还结合近似熵(approximate entropy, ApEn)分析说明利用ICA去除干扰后有助于恢复和保持原始EEG信号的非线性特征。研究结果表明ICA方法在生物医学信号处理中具有潜在的重要应用价值，值得深入研究和推广。
Blink artifacts and power noise are constantly found in EEG signals, whose acquisition and analysis can be strongly influenced by them. By comparing the efficiencies of two ICA algorithms—Infomax-ICA、Extended-Infomax-ICA and SVD methods in extracting blink artifacts and power noise in the EEG signals, ICA algorithms are insensitive to disturbance in the conditions of low signal-noise-ratio, while the commonly used SVD method does not do so well. And ICA algorithms have a strong robustness in processing non-stationary signals. Though blink slow waves can be extracted by infomax algorithm, power noise is unlikely to be removed by it. Therefore, Extended-Infomax ICA algorithm should be used. In this paper, by applying Extended-Infomax algorithms, blink artifacts and power noise contained in the 16-channel EEG signals of one Alzheimer-disease patient were removed successfully (the lowest signal-noise-ratio for power noise can be －40dB.
毕卡诗、高扬、万柏坤、朱欣、尹胜琴、杨建刚
神经病学、精神病学基础医学生物科学研究方法、生物科学研究技术
脑电眼动伪差工频干扰独立分量分析(ICA)最大熵(Infomax)奇异值分解近似熵
EEGblink artifactspower noiseIndependent Component Analysis(ICA)Infomax-ICASVDapproximation entropy(ApEn)
毕卡诗,高扬,万柏坤,朱欣,尹胜琴,杨建刚.ICA去除EEG中眼动伪差和工频干扰方法研究[EB/OL].(2003-10-26)[2025-08-02].http://www.paper.edu.cn/releasepaper/content/200310-10.点此复制
评论