Prediction of intrauterine pressure waveform from transabdominal electrohysterography

Objective. We investigated the ability to non-invasively obtain uterine electromyograms (EMG; electrohysterogram (EHG)) to predict the intrauterine pressure (IUP) waveform. Study design. Patients enrolled in a term labor study of trans-abdominal electrical activity, who underwent IUP monitoring for obstetric indication, were included in this study (n = 14). EHG signals were processed and prediction filters trained against the IUP from the same patient. Sixty-eight 10-minute traces each of EHG and IUP waveforms were reviewed (Montevideo unit (MVU) calculation) by two experienced labor nurses and one obstetrician, blinded to patient and EHG vs. IUP output. In addition, area under the contraction curve (AUC), contraction detection and duration were compared. Results. EHG-derived MVUs statistically correlated with IUP (r = 0.795; p < 0.0001), but mathematically differed by 17±20% with 83% of EHG MVUs underestimating the IUP. The coefficients of variation between the two methods were relatively high (∼20%), and these could not be substantially corrected via calibration. AUC differed by 8±19% with 69% of EHG-derived AUC underestimating the IUP. EHG detected 98% of 362 IUP contractions, with 8% over-detection. Contraction duration was similar: 56.4±11.9 s vs. 55.7±13.0 s, for IUP and EHG, respectively. Conclusion. EHG-based contraction monitoring approximates IUP monitoring for contraction detection and duration. The correlation with contraction height (MVU) and AUC is much weaker with the investigated signal-processing algorithm.