Estimate MECG from abdominal ECG signals using extended Kalman RTS smoother

Abstract

Based on a modified nonlinear dynamic ECG model, this paper presents an extended Kalman Rauch-Tung-Strebel (RTS) smoother to estimate the maternal ECG (MECG). The MECG is the predominant interference in the estimation of fetal heart rate (FHR) from abdominal ECG signals, by which, the obstetricians can determine whether the fetus is in a state of distress. For the presented smoother, an automatic parameter selection method is offered to estimate ECG signal model parameters from abdominal ECG signals themselves. Performance analysis based on both synthetic and realistic abdominal ECG signals demonstrates that the extended Kalman RTS smoother is more accurate than conventional extended Kalman filter and outperforms the other methods, such as wavelet denoising (WD) or blind source separation (BSS), due to its low computational complexity and simpler lead configuration. The presented extended Kalman RTS smoother could be applied to long-term at-home monitoring of fetal well-being.