Time-varying analysis of functional connectivity

The brain is a complex, dynamic structure that continuously processes and synchronizes responses to both internal and external stimuli across multiple time scales. Till 15 years ago, most functional magnetic resonance imaging (fMRI) research concentrated on assessment of resting state functional connectivity (RS FC) using data acquired over the entire scan session, implicitly assuming that FC remains static across time. However, it was soon realized that this strategy led to loss of critical information regarding the underlying neural fluctuations and interactions, which changed at the timescale of seconds to minutes. Time-varying functional connectivity (TVFC) opened up avenues to study such temporal variability. Sliding-window analysis is the most widely used approach to estimate TVFC by dividing fMRI time series into short temporal segments (or windows) which are then shifted in time, resulting in time-resolved FC estimates for each window. Combining it with clustering techniques has led to the identification of recurring FC states that can aid in viewing cerebral FC as a multistable process that passes through various relatively stable FC states. Several other methods have also been described to assess TVFC. This chapter provides a comprehensive overview of TVFC, covering its estimation methods, its main applications to health, neurological and psychiatric disorders, and concludes by discussing TVFC limitations and interpretability challenges, while also suggesting future directions.