采用盲源提取和后小波滤波的胎儿心电图提取

Extraction of FECG using blind source extraction and post-wavelet de-noising

胎儿心电图能够提供有关胎儿健康的信息,帮助确定胎儿是否有疾病,在临床上具有很重要的意义。将心电图仪的电极放置到孕妇体表,获取来自母体内部的心电信号,该信号是母体心电图与胎儿心电图的混合。目前主要有两种方法可以通过处理混合信号,获得胎儿心电图。其中,采用盲源提取的方法可以用较小的运算量获得胎儿心电图,但这种方法提取的胎儿心电图容易受到噪声的污染。而在小波域进行盲源分离,可以分离出较为纯净的胎儿心电图,该方法的缺点是运算量大。提出了一种新的胎儿心电提取方法,该方法首先利用盲源提取算法从采集到的心电信号中提取出受到噪声污染的胎儿心电图,然后再利用小波滤波滤除噪声,获取较为纯净的胎儿心电图。该方法既具有较小的运算量,又可以获得较好的提取性能。合成心电数据仿真和实际胎儿心电图提取试验均验证了该方法的有效性。

Fetal electrocardiogramcan （FECG） can provide information about the health and possible diseases of the fetus. So the extraction of FECG is of vital importance from clinical point of view. The ECG signals can be obtained by putting electrodes on maternal body and the signals are the mixture of maternal ECG （MECG） and FECG. Through processing the mixture signals, the FECG can be extracted using two methods at present. The first simple method is blind source extraction （BSE）, which needs less computation, but the extracted FECG is corrupted by noise easily. And the second method is blind source separation in wavelet domain, which means the BSS algorithm is performed with the wavelet coefficients of the mixture signals. This method has better performance but needs more complicated computation. A new fetal ECG extraction scheme is proposed within which the fetal ECG corrupted by noise is extracted using BSE and then the extracted signal is filtered in wavelet domain, and clearer FECG signal is obtained. This scheme can get good extraction performance with less computation, which is confirmed by computer simulation on artificial data and experiments on real-world fetal ECG data.