Investigating the spatial and temporal interactions in resting-state fMRI with total activation

Resting-state functional magnetic resonance imaging (fMRI) has become an important tool to study the spontaneous brain fluctuations. Especially, in terms of network analysis, the intrinsic brain activations have been shown to exhibit some characteristic spatial patterns referred to as resting-state networks. These distinct networks, which are mostly distinguished by their spatial distribution, have been proven to be reproducible via different fMRI data analysis techniques. However, the spatial and temporal interactions in the resting-state fMRI data are merely investigated. It is necessary to investigate the temporal evolution of the “bursts” of activations or the “switching” of different (probably overlapping) networks for better understanding of brain's intrinsic organization. Recently, we have proposed total activation (TA) which reveals the “underlying” activity-patterns in fMRI without incorporating any prior model of the events' onsets and durations. TA is cast as a spatio-temporal regularization where it promotes the sparsity of the innovation signals, which delineate the onsets and offsets of transient activations in each voxel. In this work, we study the temporal and spatial interactions of the synchronized “burst” of activations captured by TA.