Modulation of temporally coherent brain networks estimated using ICA at rest and during cognitive tasks

Abstract

Brain regions which exhibit temporally coherent fluctuations, have been increasingly studied using functional magnetic resonance imaging (fMRI). Such networks are often identified in the context of an fMRI scan collected during rest (and thus are called "resting state networks"); however, they are also present during (and modulated by) the performance of a cognitive task. In this article, we will refer to such networks as temporally coherent networks (TCNs). Although there is still some debate over the physiological source of these fluctuations, TCNs are being studied in a variety of ways. Recent studies have examined ways TCNs can be used to identify patterns associated with various brain disorders (e.g. schizophrenia, autism or Alzheimer's disease). Independent component analysis (ICA) is one method being used to identify TCNs. ICA is a data driven approach which is especially useful for decomposing activation during complex cognitive tasks where multiple operations occur simultaneously. In this article we review recent TCN studies with emphasis on those that use ICA. We also present new results showing that TCNs are robust, and can be consistently identified at rest and during performance of a cognitive task in healthy individuals and in patients with schizophrenia. In addition, multiple TCNs show temporal and spatial modulation during the cognitive task versus rest. In summary, TCNs show considerable promise as potential imaging biological markers of brain diseases, though each network needs to be studied in more detail.

Figure 1

TCNs identified for the auditory oddball task and during the resting state: TCNs identified for AOD (left side of each column) and the most spatially correlated component for REST (right side of each column). Each TCN was entered into a 1‐sample t test and is thresholded at P < 1e‐5 (corrected for multiple comparisions using the family wise error approach implemented in the SPM5 software). Four slices from each TCN are shown.

Figure 2

Differences in spectral power for controls versus patients: TCN timecourses for each subject were divided into six frequency bins (the upper range is plotted on the figure, hence the first bin is from [0–0.03 Hz.). For each bin the controls were compared to the patients using a two‐sample t test. Each TCN is plotted in a different color. Positive bars indicate frequency bins where controls are greater than patients, negative bars indicate frequency bins where patients are greater than controls. The overall pattern for all TCNs is that controls show more low frequency power and patients show more high frequency power.

Figure 3

Spatial modulation of TCNs in patients and controls for auditory oddball versus rest: Each paired TCN for AOD and REST were compared by entering the single‐subject spatial maps into a voxelwise two‐sample t test (thresholded at P < 0.05, FDR corrected). Thus for a given voxel a positive value means that the auditory oddball TCN had a larger value at that voxel than the rest TCN. Separate comparisons were performed for healthy controls (left side of each column) and patients (right side of each column).

Figure 4

Patient versus control differences between auditory oddball and rest: A direct comparison of patient versus control differences is performed by subtracting each paired TCN for AOD and REST, then entering these into a two‐sample t‐test for patients versus controls (e.g. [(AOD‐REST)_HC−(AOD‐REST)_SZ]). Results are thresholded at P < 0.05 FDR corrected.