Linear and non-linear analysis of uterine contraction signals obtained with tocodynamometry in prediction of operative vaginal delivery

J Perinat Med. 2017 Apr 1;45(3):327-332. doi: 10.1515/jpm-2016-0036.

Abstract

Objective: The aim of this study was to explore whether linear and non-linear analysis of uterine contraction (UC) signals obtained with external tocodynamometry can predict operative vaginal delivery (OVD).

Materials and methods: The last 2 h before delivery (H1 and H2) of 55 UC recordings acquired with external tocodynamometry in the labour ward of a tertiary care hospital were analysed. Signal processing involved the quantification of UCs/segment (UCN), and the linear and non-linear indices: Sample Entropy (SampEn) measuring signal irregularity; interval index (II) measuring signal variability, both of which may be associated with uterine muscle fatigue, and high frequency (HF), associated with maternal breathing movements. Thirty-two women had normal deliveries and 23 OVDs. Statistical inference was performed using 95% confidence intervals (95% CIs) for the median, and areas under the receiver operating curves (auROCs), with univariate and bivariate analyses.

Results: A significant association was found between maternal body mass index (BMI) and UC signal quality in H1, with moderate/poor signal quality being more frequent with higher maternal BMI. There was an overall increase in contraction frequency (UCN), signal regularity (SampEn), signal variability (II), and maternal breathing (HF) from H1 to H2. The OVD group exhibited significantly higher values of signal irregularity and variability (SampEn and II) in H1, and higher contraction frequency (UCN) and maternal breathing (HF) in H2. Modest auROCs were obtained with these indices in the discrimination between normal and OVDs.

Conclusions: The results of this exploratory study suggest that analysis of UC signals obtained with tocodynamometry, using linear and non-linear indices associated with muscle fatigue and maternal breathing, identifies significant changes occurring during labour, and differences between normal and OVDs, but their discriminative capacity between the two types of delivery is modest. Further refinement of this analysis is needed before it may be clinically useful.

Keywords: Cardiotocography; digital signal processing; entropy; operative vaginal delivery (OVD); spectral analysis.

Publication types

  • Observational Study

MeSH terms

  • Adult
  • Delivery, Obstetric*
  • Female
  • Humans
  • Infant, Newborn
  • Linear Models
  • Male
  • Nonlinear Dynamics
  • Portugal
  • Predictive Value of Tests
  • Pregnancy
  • Signal Processing, Computer-Assisted
  • Uterine Contraction*
  • Uterine Monitoring / statistics & numerical data*
  • Young Adult