Comparison of methods for evaluating signal synchronization and direction: Application to uterine EMG signals

Over the last decades several techniques have been developed to analyze interactions in multivariate dynamic systems. Several measures have been proposed to detect the strength and direction of relationships between time series and in physiological data. In this paper we use three methods that are widely used in connectivity and directionality analysis: Nonlinear correlation coefficient (h 2 ), general synchronization (H) and Granger causality (GC). The performances of all methods are tested on two-dimensional coupled synthetic nonlinear Rossler model. They are then applied to a dataset of contractions of real uterine EMG to study the evolution of synchronization with week before labor. Finally h 2 is used to detect the connectivity and direction between all the signal channels. The results on synthetic signal show a slight superiority of h 2 . The results obtained on real signals are encouraging and may be used for solving open questions on the directionality of EHG and may provide a manner of finding their source loci.