Assessment of coupling between trans-abdominally acquired fetal ECG and uterine activity by bivariate phase-rectified...

Couplings between uterine contractions (UC) and fetal heart rate (fHR) provide important information on fetal condition during labor. At present, couplings between UC and fHR are assessed by visual analysis and interpretation of cardiotocography. The application of computerized approaches is restricted due to the non-stationarity of the signal, missing data and noise, typical for fHR. Herein, we propose a novel approach to assess couplings between UC and fHR, based on a signal-processing algorithm termed bivariate phase-rectified signal averaging (BPRSA).

Electrohysterogram (EHG) and fetal electrocardiogram (fECG) were recorded non-invasively by a trans-abdominal device in 73 women at term with uneventful singleton pregnancy during the first stage of labor. Coupling between UC and fHR was analyzed by BPRSA and by conventional cross power spectral density analysis (CPSD). For both methods, degree of coupling was assessed by the maximum coefficient of coherence (CPRSA and CRAW, respectively) in the UC frequency domain. Coherence values greater than 0.50 were consider significant. CPRSA and CRAW were compared by Wilcoxon test.

At visual inspection BPRSA analysis identified coupled periodicities in 86.3% (63/73) of the cases. 11/73 (15%) cases were excluded from further analysis because no 30 minutes of fECG recording without signal loss was available for spectral analysis. Significant coupling was found in 90.3% (56/62) of the cases analyzed by BPRSA, and in 24.2% (15/62) of the cases analyzed by CPSD, respectively. The difference between median value of CPRSA and CRAW was highly significant (0.79 [IQR 0.69-0.90] and 0.29 [IQR 0.17-0.47], respectively; p<0.0001).

BPRSA is a novel computer-based approach that can be reliably applied to trans-abdominally acquired EHG-fECG. It allows the assessment of correlations between UC and fHR patterns in the majority of labors, overcoming the limitations of non-stationarity and artifacts. Compared to standard techniques of cross-correlations, such as CPSD, BPRSA is significantly superior.