Fast computation of multivariate synchrony index in sliding windows: application to cardiac neurons

Abstract

Multielectrode array neuronal recordings in atrial ganglionated plexi are characterized by low firing rates, marked non-stationarity, interplay with the cardiovascular and pulmonary systems and artifacts generated by myocardial activity, which creates challenges very different from brain recordings. To explore population dynamics of intrinsic cardiac neurons, a jitter-based synchrony index has been defined to quantify pairwise synchrony between neurons. In this paper, we extend this synchrony index to multiple time series in order to monitor global (multivariate) synchrony. Numerical techniques are developed to efficiently compute synchrony indices and their statistical significance in a large number of time windows. A scaletime graphical representation is proposed to visualize synchrony in sliding windows of varying lengths. This approach is validated in synthetic time series and in experimental data sets recorded in 11 dogs. Results show the ability of the method to monitor synchrony over time in neuron populations, between neurons and the cardiopulmonary system and between neuron firing and electrical stimulation. These tools will facilitate the exploration and robust quantitative analysis of multiple-hour recordings in cardiac ganglionated plexi to efficiently identify relevant periods of activity in relation to physiological or external stimuli and cardiac arrhythmia.