基于独立分量分析去除脑电中眨眼和水平扫视的伪迹  

Removal of Blink and Saccade Artifact in EEG Recordings with Independent Component Analysis

作　　者：董洁[1] 王涛[1] 张爱桃[1] 

机构地区：[1]南方医科大学生物医学工程学院,广东广州510515

出　　处：《航天医学与医学工程》2011年第2期122-127,共6页Space Medicine & Medical Engineering

基　　金：国家自然科学基金项目(60771035)

摘　　要：目的利用独立分量分析方法(ICA)将混合在观测信号中相互独立的源信号分离出来。方法记录3个正常人自然眨眼和水平扫视条件下7道脑电信号和2道眼电信号,选取7道脑电信号进行处理,2道眼电信号用来指示干扰源的情况。使用扩展相似对角化算法(JADE)将脑电信号分解成多个独立分量,同时利用伪迹脑地形图特征,判断出与眼电伪迹相关分量并将其去除。结果存在于前额电极的眼电干扰被消除,同时其他电极上的信号细节成分较好地保留下来。独立分量分析方法成功去除了脑电信号中的眼电伪迹。结论本文中算法可以用于实现脑电信号中眨眼和水平扫视干扰的去除,其中对前者的去除更加有效。Objective To remove ocular artifacts in EEG by using independent component analysis（ICA） with joint approximate diagonalization of eigenmatrice（JADE） algorithm.Methods The nine channels of EEGs were recorded from three young healthy subjects with additional VEOG and HEOG channels in blinking and saccade conditions respectively.Seven channel EEG records were selected for artifacts removal tests with two EOG records indicating the possible interference.The raw EEG signals were separated into several independent components by JADE algorithm.According to these scalp topography,the components being judged as eye artifacts were eliminated.Results By using the time domain and frequency domain analysis of the raw data and the result data,the peaks caused by eye movements on the frontal channels were mostly removed and the original details of EEG information were preserved.Conclusion ICA method can remove more artifacts produced by eye blinking than that by saccade.This study demonstrates that JADE algorithm may be an effective tool in correcting EOG interference with multichannel EEG recordings.

关 键 词：独立分量分析 脑电图 眼电伪迹 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] R318[自动化与计算机技术—计算机科学与技术]