Nonlinear dynamic tools for characterizing abdominal electromyographic signals before and during labour

Abstract

This paper presents evidence that electromyographic signals during healthy uterine contractions may have a fractal temporal structure. Healthy uterine contractions start from the 16th week of the human pregnancy. These contractions become more and more frequent and increase in strength up to the end of pregnancy. Here we investigate the possibility that time series generated by labour physiological control systems may be members of a special class of complex processes which are noise driven and may require more than one exponent to characterize their fractal scaling properties. We use a simplified version of the Hurst analysis algorithm to detect fractility in abdominal electromyographic signals (AEMG) and quadratic Volterra structures to estimate nonlinearity in them. We uncover a loss of chaoticity in very early segments of uterine contractions for typical cases of failure to progress in the first stage of labour and ending with surgical procedure (Caesarean section). Results are shown for representatives of the following groups: (1) healthy pregnancy, spontaneous labour and parturition, (2) full term at 40 weeks, prolonged labour (arbitrarily defined as lasting more than 12 h) and ending with Caesarean delivery, and (3) preterm spontaneous labour (less than 37 weeks).

An Uterine Electromyographic Activity as a Measure of Labor Progression

“…& Russell K.L., 1989;Albers et al, 1996). The mean frequency of contractions over the active phase of labor is four contractions in 10 min, rising to five in 10 min in the late first stage (Sabry-Rizk et al, 2000). The duration of contractions remains remarkably constant, between 70 s and 90 s. The second stage begins with complete cervical dilatation and ends with the delivery of the fetus (Chard & Grudzinskas, 1994).…”

“…Precisely, by studying its fractal behavior it was shown that the size of the nucleus increases with cervical dilatation until it takes over the entire phase-space. The weak labor contractions, typical for failure in progress in the first stage of labor, showed predominantly periodic structures with or without small-size nuclei (Sabry-Rizk et al 2000). Maner et al used wavelet transform on the bursts' traces and then calculated fractal dimension of the resulting transformed EMG burst-trace.…”

Virtues and vices of source separation using linear independent component analysis for blind source separation of non-linearly coupled and synchronised fetal and mother ECGs

“…Even if this is justified as first approximation, the nonlinearities which characterise individual mother and fetal cardioelectrical activities [22], [23] will interact during their propagation through various body layers and mix with motion artefact, before they are finally picked up as the ECG signals by electrodes on the skin surface. It is important to realise that the cardiac signals have to penetrate through a complex system experiencing various effects and there is evidence of spectral tuning between the fetal heartbeat and uterus contractions [26], [27]. However, only evidence of quadratic coupling between the mother and fetal ECGs will be given in this paper due to lack of space.…”

A new class of non-linear, multi-dimensional structures for long-term dynamic modelling of chaotic systems

“…Recently, artificial neural networks (ANN'S) based architectures have been proposed in the litcraturc [4], which are based on the theory of functional expansions of non-linear, nonstationary Volterra series. However, the need for improving the accuracy of mapping (the dynamicreconstruction problem) in applications involving multkfractal biological dynamical systems such as nine-dimensional lahour contractions [ 5 ] and heart-rate [ 6 ] has sparked great interest in attempting to develop a new family of adaptive temporal one-or multiple-step embedded Volterra predictors which can be used to examine the short-term prediction properties of the attractor with sufficient memory to include relevant past non-linear correlation behavioural pattem of a finite signal sample. Characterizing such complex chaotic systems by state-space analysis of observable data sample, provides a framework for designing longterm predictors.…”

“…Section 2 describes the embedding of a conventional Volterra architecture to create a family of dynamic-based one-step Volterra architectures that capture the underlying local dynamics and subsequently give superior performance to mono-fractal chaotic systems [ 5 ] , [8]. Section 3 discusses the results obtained from applying the regularized RBF networks and the novel embedded multidimensional weighted structures.…”