Detection of Transient Inter-regional Coupling in fMRI Time Series: A New Method Combining Inter-subjects Synchronization and Cluster-Analyses

We present a new method for the analysis of fMRI time series. The aim is to identify functionally-relevant transient "bursts" of inter-regional coupling between brain areas, using a fully data-driven approach. We use inter-subjects synchronization (i.e. correlation between time series of different subjects who are presented with the same sensory input) to isolate relevant transients in the fMRI time series. Next, we apply a first cluster analysis to group together areas that show such synchronized activity in a concurrent manner. Finally, a second cluster analysis identifies patterns of the fMRI signal that repeat consistently across the different transients. The final output of the analysis is a set of networks that show transient patterns of functionally relevant fMRI signal, consistently over specific windows of the time series. Importantly, the fMRI signal can differ between different areas belonging to the same network. This new approach is particularly suited to investigate multi-components control processes using naturalistic stimuli during fMRI.