Improved Detection of Nonlinearity in Nonstationary Signals by Piecewise Stationary Segmentation

Recently, much attention has been paid to the use of nonlinear analysis techniques for the characterization of biological signals. These signals are however often strongly non-stationary. This is in contradiction to the assumption of commonly used nonlinearity measures, which assume that the signal is stationary. We propose to use a Piecewise Stationary Segmentation (PSP) of the signals of interest, before the computation of nonlinearity measures. We show on synthetic as well as real signals (speech and uterine EMG) that the proposed piecewise stationary segmentation approach increases the accuracy of the measures by making a good trade-off between the stationary assumption and length of the analyzed segments, when compared to the classical windowing method.