De-noising the abdominal phonogram for foetal heart rate extraction: Blind source separation versus empirical filtering

This work explored the suitability of using the foetal phonocardiogram (FPCG) blindly separated from the abdominal phonogram as a source for foetal heart rate (FHR) measuring in antenatal surveillance. To this end, and working on a dataset of 15 abdominal phonograms, the FPCG was estimated by using two de-noising approaches (1) single-channel independent component analysis (SCICA) to produce FPCG_e and (2) empirical filtering to produce FPCG_g. Next, the FPCGs were further processed to collect the beat-to-beat FHR and the resulting time-series (FCTG_e and FCTG_g) were compared to the reference signal given by the abdominal ECG (FCTG_r). Results are promising, the FPCG_e gives rise to a FCTG_e that resembles FCTG_r and, most importantly, whose mean FHR value is statistically equivalent to that given by FCTG_r (p > 0.05). Thus, the mean FHR value obtained from the FPCG_e, is likely to be equivalent to the value given by the abdominal ECG, which is especially significant since the FPCG_e is retrieved from the noisy abdominal phonogram. Hence, as far as this study has gone, it can be said that, when using SCICA to de-noise the abdominal phonogram, the resulting FPCG is likely to become a useful source for FHR collection in antenatal surveillance.