结合多路ν-SVR与TFBSS的胎儿心电信号提取  

FECG signal extraction based on multichannel v-SVR combined with TFBSS

作　　者：韩亮[1] 蒲秀娟[1] 刘茜[1] 

机构地区：[1]重庆大学通信工程学院,重庆400044

出　　处：《仪器仪表学报》2015年第6期1381-1387,共7页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金面上项目(61171158);重庆市自然科学基金(cstc2012jj A10128);中央高校基本科研业务费(CDJZR12160010)项目资助

摘　　要：提出了一种结合多路v-支持向量回归机(ν-SVR)和时频盲源分离(TFBSS)的胎儿心电信号间接提取方法。利用多路ν-SVR估计各路腹壁信号中的母体心电成分并将其抑制,从而得到多路含噪声的胎儿心电成分的最优估计;然后通过TFBSS从多路含噪声的胎儿心电成分的最优估计中提取胎儿心电信号。采用临床心电数据进行实验,结果表明本文提出的方法在可视化结果和信噪比指标上均优于传统方法。Aiming at the difficulty of fetal electrocardiogram signal extraction,a novel method of fetal electrocardiogram（ FECG） signal indirect extraction is proposed,which combines multichannel v-support vector regression（ ν-SVR） and blind source separation based on time-frequency distribution（ TFBSS）. Firstly,the multichannel ν-SVR is used to obtain the optimal estimation of the maternal ECG（MECG） component in the multichannel abdominal signal. Then,the optimal estimation of the FECG containing noise is obtained by removing the MECG component from the abdominal signal. Finally,the TFBSS is used to extract the FECG from the multichannel FECG containing noise. The clinical ECG data were used to perform the experiment on fetal electrocardiogram signal extraction. The performance of the proposed method is assessed utilizing visual result,signal-to-noise ratio（ SNR） and time complexity. The experiment results show that the proposed method is able to extract clear FECG signal and has significant improvement in SNR compared with traditional fetal electrocardiogram signal extraction methods. The experiment results verify the effectiveness of the proposed method.

关 键 词：胎儿心电信号 v-支持向量回归机 时频盲源分离 

分 类 号：TN911.72[电子电信—通信与信息系统] TH701[电子电信—信息与通信工程]