非侵入式胎儿心电信号的R波检测方法

Non-invasive R-wave detection of fetal electrocardiogram

采用非侵入式方法从母体腹部采集的胎儿心电信号(FECG),由于信号微弱并受到母体心电信号和噪声的干扰,要准确识别出FECG的R波非常困难.本文通过对从母体腹部表面采集的信号进行二进小波变换,并用模极大值法确定信号奇异值点的位置,根据母体与胎儿心电在模极大值上的差异,识别出胎儿心电信号的R波.此方法在MIT Physio Net/Cin C 2013挑战竞赛公布的标准数据进行了实验,并参加了SET-B数据集的挑战,根据MIT竞赛平台公布的结果,此方法从母体腹部采集的单通道信号中识别胎儿心电信号R波的准确率达到93.8%.

It is difficult to detect the R-wave of a fetal electrocardiogram（ECG） from abdominal ECG recordings, because the FECG is a kind of weak signal and is interfered by mother＇s ECG. In the paper, the binary wavelet transform is used to calculate the wavelet coefficients in five scales of the signal, and moreover, the modulus maximum is used to find the singular value points, then the R-wave of a FECG is detected according to different modulus maxima of them. The method is evaluated on the signal datasets obtained from 2013 Physio Net/Cin C Challenge, which are collected from maternal abdominal surface through four channels simultaneously. The results show that the R-wave detection accuracy rate is 93.8%.