心电信号的处理

Biomedicalsignalprocessing1NankaiUniversity,CYLI,2020/2/2Ch4:心电信号的处理心电图的产生,心电图处理的基本思路,时间上:动态和静态,空间上:心肌电特性的空间离散度,心电图处理和分析的发展,Biomedicalsignalprocessing2NankaiUniversity,CYLI,2020/2/2心电信号的形成（1）心肌细胞的动作电位心肌细胞除极和复极的电生理现象,极化状态(polarised):静息电位(restingpotential),动作电位(actionpotential,AP):除极(depolarisation)和复极(repolarisation),mV-90-600+200心室肌细胞动作电位示意图1234300msBiomedicalsignalprocessing3NankaiUniversity,CYLI,2020/2/2心电信号的形成（2）电兴奋的传导(conductionorspreadofelectricalexcitation)窦房结:心脏的起搏兴奋点,其细胞自发产生50-100次/分的可传导AP,心房的传导:心房传导束--右心房--左心房传导系统的传播:房室束--希氏束--左、右分支--普金野（Purkinje）纤维网-心室肌,电兴奋通过Purkinje网使心室肌细胞兴奋，Biomedicalsignalprocessing4NankaiUniversity,CYLI,2020/2/2心脏传导系统Biomedicalsignalprocessing5NankaiUniversity,CYLI,2020/2/2Englishterms窦房结:sinusnode,心房:atrium,心室:ventricle,房室结:atrioventricularnode(junction),希氏束:bundleofHis,普金野纤维:Pukinjefibres,心脏表面(心外膜):epicardium,心内膜：endocardium体表：bodysurfaceBiomedicalsignalprocessing6NankaiUniversity,CYLI,2020/2/2心电信号的形成（3）心电图(electrocardiogram,ECG)的产生电流源：每个心肌细胞的除极和复极过程等效于一个偶极子层（dipolelayer）容积导体(volumeconductor):人体组织是导电的，看作是一个容积导体，心电向量:所有心肌细胞的偶极子场的向量和,心电图：所有心肌细胞的偶极子场在容积导体内产生电场，从而有电位差产生，即心电图。体表心电图,心外膜、心内膜电图,希氏束电图，Biomedicalsignalprocessing7NankaiUniversity,CYLI,2020/2/2典型心电信号波形P,QRS,ST,TS-T段QRSTPQTBiomedicalsignalprocessing8NankaiUniversity,CYLI,2020/2/2心电图的记录(RecordingofECG)标准导联,肢体导联,胸导联,体表多部位标测(bodysurfacemapping),32-512通道的数据采集系统(multi-channeldataacquisitionsystem),,电极背心,(electrodevest)心脏表面的多部位标测(epicardialmapping),多通道的数据采集系统电极套(electrodesock)心脏内膜的多部位标测(endocardialmapping),多通道的数据采集系统导管(catheter)和伞电极Biomedicalsignalprocessing9NankaiUniversity,CYLI,2020/2/2BodysurfacemappingBiomedicalsignalprocessing10NankaiUniversity,CYLI,2020/2/2EpicardialelectrodesockBiomedicalsignalprocessing11NankaiUniversity,CYLI,2020/2/2标准导联(1)12标准导联(standardleads)标准I，II，III肢体导联(bipolarlimbleads),IIIIIIRFLR:rightarmL:leftarmF:leftfootI=EL-ER,II=EF-ER,III=EF-ELBiomedicalsignalprocessing12NankaiUniversity,CYLI,2020/2/2标准导联(2)加压肢体导联标准aVR,aVL,aVF(augmentedunipolarlimbleads),LWilsonterminal:C点的电位,aVL和aVF的连接相似R/2RRRFCaVR+-Biomedicalsignalprocessing13NankaiUniversity,CYLI,2020/2/2标准导联(3)单极胸导联V1-V6(unipolarprecordialleads),precordial:chestwall,中心电端(Wilsonterminal)V1-V6,胸前电极分别与中心电端的电位差,Biomedicalsignalprocessing14NankaiUniversity,CYLI,2020/2/2心电信号的畸变心脏的病变：传导阻滞，早搏，室颤、房颤,心肌缺血、梗塞等,QRS变宽,ST段位移出现,心率变化等,来自心脏外的干扰信号：50Hz工频干扰肌电干扰，10-300Hz,呼吸的干扰，使基线漂移加剧，Biomedicalsignalprocessing15NankaiUniversity,CYLI,2020/2/2心电信号的预处理抑制工频干扰，基线纠漂，Biomedicalsignalprocessing16NankaiUniversity,CYLI,2020/2/2噪声抑制和基线漂(detrending)低通滤波器可以滤掉心电信号中的肌电信号的高频干扰，抑制工频干扰自适应滤波抑制工频干扰，基线漂移的纠正抵消法纠漂，基线纠漂滤波器，0.7Hz的高通截止频率，Biomedicalsignalprocessing17NankaiUniversity,CYLI,2020/2/2移动平滑滤波器(低通滤波)1(2)2,(1)1,(0);1)(2,2)(,1(0))]2()(2)([12)(1)(2)(21211111)(1(1)1,(0);1)(,1(0))]()([11)()(111)()()()...(1)(aaaNNbNNbNbNnxNnxnxNnynynyZZZZNZZNzHaaNNbNbNnxnxNnyny-z-zNzHknxnyZZZzHBiomedicalsignalprocessing18NankaiUniversity,CYLI,2020/2/2H(z)和H(z)*H(z)的比较H(z):旁瓣太大,13dB;相频特性虽在通带内保持线性,但在进入阻带后有突变,有可能造成心电信号的高频相位失真;H(z)*H(z):旁瓣有较大的衰减,26.7dB,有真正的线性相位;Biomedicalsignalprocessing19NankaiUniversity,CYLI,2020/2/2移动平滑滤波器(N=8)Fs=1000HzN=8,H(z),-3dB,55Hz,thefirstsidepeak:-13dB,nonlinearphaseangleH(z)*H(z),-3dB,55Hz,thefirstsidepeak:-26.7dB,linearphaseangleBiomedicalsignalprocessing20NankaiUniversity,CYLI,2020/2/2移动平滑滤波器(N=33)H(z):-3dB:0.08Hz,thefirstsidepeak:-13dBH(z)*H(z):-3dB:0.08Hz,thefirstsidepeak:-26.7dBBiomedicalsignalprocessing21NankaiUniversity,CYLI,2020/2/2移动平滑滤波器的效果Biomedicalsignalprocessing22NankaiUniversity,CYLI,2020/2/2MA的H(z)和H(z)*H(z)随N的变化00.20.40.60.81-60-50-40-30-20-100MAH(z)pi00.20.40.60.81-4-3-2-1012H(z)Angle00.20.40.60.81-60-50-40-30-20-100H(z)*H(z)pi00.20.40.60.81-120-100-80-60-40-200H*HAngleN=8N=13N=8N=13Biomedicalsignalprocessing23NankaiUniversity,CYLI,2020/2/2自适应滤波根据输入信号自动调节滤波器的参数,使其性能指标最优化，适用于对信号和噪声无先验知识(频谱)的或非平稳信号,基本结构:滤波器、优化指标算法、滤波器参数修改算法，滤波器：FIR,IIR，优化准则：信噪比最高、输出误差均方差最小，参数修改：递归、非递归，Biomedicalsignalprocessing24NankaiUniversity,CYLI,2020/2/2y(t)自适应消噪声+滤波器修改参数信号源噪声源x(t)=s(t)+n0(t)x(t)+-n1(t)n0(t):噪声源（如50Hz工频信号），n1(t)：通过某一未知网络的同一噪声源，n0’(t)：滤波器输出的估值，x(t)=s(t)+n0(t)-y(t)=s(t)+n0(t)-n0’(t)要使得x(t)在最小均方差意义上与信号s(t)最佳匹配，的最佳估值是最小时当s(t)x(t),})]()({[})]()({[)]([)]([2'002'0022tntnEtntnEtsEtxEBiomedicalsignalprocessing25NankaiUniversity,CYLI,2020/2/2基线纠漂抵消法：从信号中减去基线的估计值,心电基线漂移可看成某种超低频干扰信号,x(n)=s(n)+w(n),用w’(n)来逼近w(n),并从s(n)中减去,s(n)=x(n)-w’(n),PQ段为基线，常以PQ段的中点作为基准点，用多项式拟合基线的估计函数，如最简单的线性拟合、三次样条函数拟合等，x(n)=s(n)+w(n)基线估计+-s(n)w’(n)Biomedicalsignalprocessing26NankaiUniversity,CYLI,2020/2/2基线纠漂滤波器基线纠漂滤波器(HP)高通滤波器，高通截止频率：0.7Hz，即以40次/分为心脏搏动过缓的下限，基线纠漂及50Hz陷波滤波器,理想的心电信号预处理滤波器基线纠漂+50Hz限波,可用频率抽样法设计这种滤波器,fH(f)50100150fkBiomedicalsignalprocessing27NankaiUniversity,CYLI,2020/2/2纠漂和50Hz陷波的效果Biomedicalsignalprocessing28NankaiUniversity,CYLI,2020/2/2整系数基线纠漂滤波器梳状滤波器的数学表示;8451225624256212164*641)(11641)()4()]256()([641)(zzzzzHzzzHnyn