A Sliding Time-Window ICA Reveals Spatial Variability of the Default Mode Network in Time

The spatial layout of the brain's default mode network (DMN) fluctuates over time, even when a person is at rest. Using a sliding time window analysis of functional connectivity MRI data from 12 participants, the authors found that the spatial similarity of successive DMN maps varied at low frequencies and followed a 1/f power spectral pattern. A single voxel appeared within a group DMN map in at most 82% of time windows, indicating marked temporal variation. Incidental connectivity increased centrifugally toward brain areas outside the DMN core. The method, SliTICA, captures this spatial variance and may provide a more complete measure of a network's functional activity.

Study at a glance

Abstract

Recent evidence on resting-state networks in functional (connectivity) magnetic resonance imaging (fcMRI) suggests that there may be significant spatial variability of activity foci over time. This study used a sliding time window approach with the spatial domain-independent component analysis (SliTICA) to detect spatial maps of resting-state networks over time. The study hypothesis was that the spatial distribution of a functionally connected network would present marked variability over time. The spatial stability of successive sliding-window maps of the default mode network (DMN) from fcMRI data of 12 participants imaged in the resting state was analyzed. Control measures support previous findings on the stability of independent component analysis in measuring sliding-window sources accurately. The spatial similarity of successive DMN maps varied over time at low frequencies and presented a 1/f power spectral pattern. SliTICA maps show marked temporal variation within the DMN; a single voxel was detected inside a group DMN map in maximally 82% of time windows. Mapping of incidental connectivity reveals centrifugally increasing connectivity to the brain cortex outside the DMN core areas. In conclusion, SliTICA shows marked spatial variance of DMN activity in time, which may offer a more comprehensive measurement of the overall functional activity of a network.