Test-retest reproducibility of the default-mode network in healthy individuals

Abstract

Independent component analysis (ICA) of functional magnetic resonance imaging (fMRI) time-series reveals distinct coactivation patterns in the resting brain representing spatially coherent spontaneous fluctuations of the fMRI signal. Among these patterns, the so-called default-mode network (DMN) has been attributed to the ongoing mental activity of the brain during wakeful resting state. Studies suggest that many neuropsychiatric diseases disconnect brain areas belonging to the DMN. The potential use of the DMN as functional imaging marker for individuals at risk for these diseases, however, requires that the components of the DMN are reproducible over time in healthy individuals. In this study, we assessed the reproducibility of the DMN components within and between imaging sessions in 18 healthy young subjects (mean age, 27.5 years) who were scanned three times with two resting state scans during each session at 3.0 T field strength. Statistical analysis of fMRI time-series was done using ICA implemented with BrainVoyager QX. At all three sessions the essential components of the DMN could be identified in each individual. Spatial extent of DMN activity and size of overlap within and between sessions were most reproducible for the anterior and posterior cingulate gyrus. The degree of reproducibility of the DMN agrees with the degree of reproducibility found with motor paradigms. We conclude that DMN coactivation patterns are reproducible in healthy young subjects. Therefore, these data can serve as basis to further explore the effects of aging and neuropsychiatric diseases on the DMN of the brain.

Figure 1

Calculation of R _size and R _overlap. The figure shows the extent of the DMN regions in two rsEPIs (rsEPI1, light blue; rsEPI2, dark blue). Overlapping voxels are marked in red.

Figure 2

Coactivated areas contributing to the DMN of the brain averaged over subjects in session 1 (a), session 2 (b), and session 3 (c). (averaged anatomical data set, Talairach coordinates, x = −4, y = −53, z = 24).

Figure 3

Examples for intersession (a) and intrasession (b) overlap in two subjects. In (a), the R _size for the ACC and PCC were 0.94 and 0.96, respectively. The PCC showed a reasonable R _overlap of 0.65, the ACC revealed a R _overlap of only 0.09. In (b), the ACC and PCC show an R _size of 0.79 and 0.92, respectively, and a R _overlap of 0.71 and 0.82, respectively.