去除脑电信号眨眼伪差的独立成分时域相关算法  

A Method for Removing Eye-Blink Artifacts from EEG Signals by Temporal Correlation of Independent Component Analysis

作　　者：盛恒松[1] 田洪君 郑崇勋[1] 徐进[1] 

机构地区：[1]西安交通大学生物医学信息工程教育部重点实验室,西安710049 [2]南京丰盛超导技术有限公司,南京211113

出　　处：《西安交通大学学报》2013年第10期127-131,共5页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基(31271061;81271659)

摘　　要：针对脑电信号易受眨眼动作干扰的问题,提出了一个自动地去除眨眼伪差的独立成分时域相关算法。该算法使用独立成分进行分析,并分解脑电信号,然后利用眨眼伪差独立成分与某些导联的脑电信号之间在时域存在较大相关性的特点,计算每个独立成分与前额附近的5个导联(Fp1,Fp2,F3,F4,Fz)信号的相关值的累加值,并对该值进行排序,将具有最大值的独立成分识别为眨眼伪差独立成分,将其设置为0,最后重建干净的脑电信号。通过对脑电信号的去除伪差实验表明:眨眼伪差引起的干扰基本被消除,伪差检测算法的敏感度和特异度分别是97.7%和98.3%,同时该算法能有效保持脑电信号基本不变。A novel method is proposed to remove eye-blink artifacts from EEG signals automatically.The EEG signals are decomposed by independent component analysis （ICA）,and the features of temporal correlation between ICs and observed EEG signals from some electrodes are extracted.The sum of the correlation between each IC and EEG signal from respective electrodes Fp1,Fp2,F3,F4 and Fz is then calculated.These correlation values are sorted in descending order.The IC with the biggest correlation value in all the ICs is picked out for the eye-blink artifact component,and is finally counted as zero to reconstruct clean EEG signals.Experiments for purifying contaminated EEG signals show that eye-blink artifacts are successfully removed from EEG signals,the sensitivity and specificity of the artifacts detection algorithm get to 97.7％ and 98.3％,respectively,remaining the original EEG signal feature.

关 键 词：时域相关 独立成分分析 脑电 眨眼伪差 

分 类 号：R318.4[医药卫生—生物医学工程]