实时脑电信号眼电伪差去除方法的研究  

Method of Electrooculogram Artifacts Elimination in Electroencephalogram

作　　者：刘明宇[1] 王珏[1] 魏娜[1] 燕楠[1] 郑崇勋[1] 

机构地区：[1]西安交通大学生命科学与技术学院,西安710049

出　　处：《西安交通大学学报》2004年第12期1306-1309,共4页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目 (30 370 395 ) ;陕西省科学技术厅资助项目 (2 0 0 3K10 G2 4 )

摘　　要：基于信号的小波包分解方法 ,提出了一种实时性好、滤波效果佳的脑电信号眼电伪差去除的新方法 .该算法在采用小波包分析将脑电信号在频域做精细划分的基础上 ,用统计学的理论选取阈值 ,对信号的特定成分进行选择性滤波 ,并选用与眼电 (EOG)伪差相关度最大的FP1作为参考信号 ,来避免算法本身引入的新伪差 ,以达到快速有效去除眼电伪差的效果 .试验结果表明 ,相对于其他几种常用的算法 ,本算法不仅具有更好的去伪效果 ,而且运算速度也比独立分量分析 (ICA)算法快 2～ 3倍 。Based on the wavelet packet analysis, a novel method is proposed to eliminate electrooculogram (EOG)-artifact from electroencephalogram (EEG). It consists of three steps: (1) EEG decomposing based on wavelet packet; (2) EOG detection and elimination from low-frequency wavelet components; (3) Reconstruction of EEG signals using FP1 signals as reference. Ninety sets of EEG from ten subjects were processed and the results were compared with the other four conventional methods. The results show that this method is more effective on EOG artifacts elimination and the processing rate is 2-3 times faster than that of the independent component analysis (ICA)-based method.

关 键 词：脑电 眼电伪差 小波包 独立分量分析 

分 类 号：R318.04[医药卫生—生物医学工程]