Fetal Heart Rate Estimation: Adaptive Filtering Approach vs Time-Frequency Analysis

Fetal heart rate (FHR) estimation is an essential aspect of fetal heart rate observation utilized to determine fetal health condition during the antenatal period. This paper proposes two methods for FHR estimation from an abdominal electrocardiogram (ECG) signal based on adaptive filtering and time-frequency techniques. The fetal ECG signal is the main aspect of the abdominal ECG marred by other interference/ noise signals. These signals include propagated maternal ECG signal, the power-line noise and the white noise. The adaptive filtering approach is based on the least mean squares (LMS) and requires the maternal ECG signal as reference signal for the filtering processing for FHR estimation. The time-frequency analysis is based on the spectrogram time-frequency distribution (TFD), where instantaneous power (IP) and peak analyses are subsequently carried out from this TFD for FHR estimation. The adaptive filtering is made of a single step but requires an external reference signal while the time-frequency analysis involves more steps but does not require a reference signal. Results obtained showed that both methods are able to estimate the FHR of 160 beats per minute (bpm) accurately at signal-to-noise ratio (SNR) of slightly higher noise at -1dB. Results obtained showed that the proposed techniques are efficient for fetal heart monitoring and choice of technique will depend on available resources.