基于QRS波群独立建模的心电信号降噪方法  

ECG signal denoising algorithm of independent modeling based on QRS complex

作　　者：李远禄[1,2] 钱仁飞 刘宝莹 李昆 LI Yuanlu;QIAN Renfei;LIU Baoying;LI Kun(School of Automation,Nanjing University of Information Science&Technology,Nanjing 210044,China;Collaborative Innovation Centre on Atmospheric Environment and Equipment Technology,Nanjing University of Information Science&Technology,Nanjing 210044,China)

机构地区：[1]南京信息工程大学自动化学院,江苏南京210044 [2]南京信息工程大学江苏大气环境与装备技术协同创新中心,江苏南京210044

出　　处：《现代电子技术》2022年第15期51-57,共7页Modern Electronics Technique

基　　金：国家自然科学基金资助项目(61671010);江苏省自然科学基金资助项目(BK20161535);江苏省高校“青蓝工程”项目(B2018Q03)。

摘　　要：在心电信号(ECG)采集过程中,噪声的干扰将极大地影响心电图信号的分析和处理,为了提取正确的生理信息,需要消除心电图信号中存在的噪声,心电信号中QRS波群包含了大量细节信息,常用去噪方法在消除噪声的同时会模糊QRS波的细节部分。因此,提出一种基于QRS波群独立建模的心电信号降噪方法。为了有更好的降噪效果,将心电信号划分为QRS波群和低频信号两部分,分别建立去噪模型。首先,利用连续小波变换(CWT)峰检测方法提取信号的QRS波群,选择惩罚最小二乘法(PLS)对其进行处理;然后使用分段分类协同滤波方法(SCCS)处理其他部分。当添加的噪声信号为5 dB,10 dB,15 dB,20 dB的高斯白噪声时,去噪后的心电信号的平均信噪比提升为11.46 dB,10.73 dB,9.87 dB,8.53 dB,平均均方根误差为0.0033,0.0014,0.0005,0.00033,平均PRD为15.25,9.26,5.69,3.74。实验数据来源于MIT⁃BIH心律异常数据库,最后的仿真实验结果表明提出的方法可以对心电信号进行有效去噪。The interference of noise will greatly affect the analysis and processing of ECG signals in the process of signal acquisition.Therefore,it is necessary to eliminate the noise existing in the ECG signals for extraction of the correct physiological information.The QRS complex in ECG signals contains a lot of detailed information.The common denoising methods will blur the details of QRS complex while eliminating noise,so an ECG signal denoising method of independent modeling based on QRS complex is proposed.In order to get better denoising effect,the ECG signals are divided into two parts,named QRS complex and low⁃frequency signals.The denoising models are established respectively.The continuous wavelet transform(CWT)peak detection method is used to extract the QRS complex of the signals,and the penalized least squares(PLS)method is selected to process the QRS complex.And then,the segment classification cooperative smoothing(SCCS)method is used to process the remaining parts.When the noise signals added are white Gaussian noise(WGN)of 5 dB,10 dB,15 dB and 20 dB,the average signal⁃to⁃noise ratio(SNR)of the denoised ECG signals is increased to 11.46 dB,10.73 dB,9.87 dB and 8.53 dB,and their average mean square errors are 0.0033,0.0014,0.0005 and 0.00033,and their mean PRDs are 15.25,9.26,5.69 and 3.74.The experimental data in this research comes from MIT⁃BIH arrhythmia database.The final results of simulation experiments show that the proposed method can effectively denoise ECG signals.

关 键 词：心电信号降噪 QRS波群 连续小波变换 分段分类协同滤波 惩罚最小二乘 高斯白噪声 

分 类 号：TN911.7-34[电子电信—通信与信息系统] TP319.4[电子电信—信息与通信工程]