资源描述:
《ICA去除EEG中眼动伪差和工频干扰方法研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、_______________________________________________________________________________www.paper.edu.cnICA去除EEG中眼动伪差和工频干扰方法研究万柏坤,朱欣,高扬,杨建刚,毕卡诗,尹胜琴(天津大学精密仪器与光电子工程学院生物医学工程与科学仪器系,天津300072)摘要: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(IndependentComponentAnalysis,ICA)方法分离EEG中此类噪声。通过对早老性痴呆症(Alzhei
2、merdisease,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(ExtendedInfomax)等ICA算法和奇异值分解(singularvaluedecomposition,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(approximateentropy,ApEn)分析说明利用ICA去除干扰后有助于恢复
3、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。关键词:脑电(EEG);眼动伪差;工频干扰;独立分量分析(ICA);最大熵(Infomax);奇异值分解(SVD);近似熵(ApEn)4、(英文)题名、作者姓名及所在单位、摘要、关键词:TheMethodResearchofApplyingIndependentComponentAnalysisToRemoveBlinkArtifactsandPowerNoiseWANBaikun,ZHUXin,GAOYang,YANGJiangang,DhakalBikash
4、,YINShengqin(DepartmentofBiomedicalengineering&ScienceInstrument,CollegeofPrecisionInstrument&Opto-electronicsinTianjinUniversity,Tianjin300072,China)Abstract:BlinkartifactsandpowernoiseareconstantlyfoundinEEGsignals,whoseacquisitionandanalysiscanbestronglyinfluencedbythem.Bycomparingtheeffic
5、ienciesoftwoICAalgorithms—Infomax-ICA、Extended-Infomax-ICAandSVDmethodsinextractingblinkartifactsandpowernoiseintheEEGsignals,ICAalgorithmsareinsensitivetodisturbanceintheconditionsoflowsignal-noise-ratio,whilethecommonlyusedSVDmethoddoesnotdosowell.AndICAalgorithmshaveastrongrobustnessinproc
6、essingnon-stationarysignals.Thoughblinkslowwavescanbeextractedbyinfomaxalgorithm,powernoiseisunlikelytoberemovedbyit.Therefore,Extended-InfomaxICAalgorithmshouldbeused.Inthispaper,byapplyingExtended-Infomaxalgorithms,blinkartifactsandpowernoisecontainedinthe16-channelEEGsignalsofoneAlzheimer-
7、diseasepatientwereremovedsuccessfully(thelowestsignal-noise-ratioforpowernoisecanbe-40dB.Meanwhile,itisprovedbycalculatingapproximationentropy(ApEn)thatICAalgorithmscanpreservethenonlinearcharacteristicsofEEGafterremovingtheinterference.ICAha