Sleep Brain Networks and Sleep Depth during Somatosensory Stimulation

Background: The resting brain presents a spontaneous activity measured by BOLD (Blood Oxygen Level Dependent) signal that encompasses definite cerebral areas and that has initially been described as Default Mode Network (DMN). This BOLD signal activation undergoes modifications, mainly over the frontal areas, throughout sleep and states of reduced consciousness. Advanced analysis method, such as the Independent Component Analysis (ICA) permit to disentangle the diverse activations of an idling brain, leading to the depiction of Resting State Networks (RSNs), that are, different brain areas related to physiological functions (i.e., the auditory or the sensory-motor network). While we know that in wake somatosensory stimulation provokes the activation of the contralateral sensory-motor cortex (Manganotti et al., 2009), we ignore the fate of the other networks if the subject falls asleep.

Methods: We report two cases of subjects falling spontaneously asleep in the bore of the scanner during experimental sessions aimed at studying the effect of wrist electrical stimulation on BOLD signal. A multivariate analysis such ICA was used, by dividing the imaging data in several spatial patterns or independent activation maps. In detail, for each subject, spatial ICA was used for the decomposition of fMRI time-series into brain activity patterns (Brain Voyager QX software – Brain Innovation, Maastricht, The Netherlands). Thirty independent components (ICs) were estimated for each subject using the FastICA algorithm (Hyvärinen, 1999). Finally, a threshold of z scores was used for visualization of the IC maps and voxels with absolute z scores > 1.5 were considered to be IC active voxels.

Results: In the sleeping condition, we identified the persistence of the DMN, occipital, attentive and temporal networks, that großly preserve the BOLD activation areas with a slight tendency towards larger activation. In both cases, during sleep we observed an important thalamus activation concordant with the hypothesis of a thalamic gating role (Del Felice et al., 2012). One subject showed during sleep an activation of the sensory-motor area concordant with the electrical stimulation, although less diffuse than during wake: this subject presented at detailed analysis of the DMN the persistence of a frontal activation during sleep. Subject 2 showed a fading off of DMN frontal areas, accompanied by the lack of congruous activation in the somatosensory areas despite the electrical stimulus.

Conclusion: To our knowledge, these are the first reported cases of fMRI sleep recordings with a concomitant sensory stimulation. The discrepancy of findings in the two cases points to a shallower sleep stage in subject 1, with the persistence of a BOLD signal over the sensory-motor area contralateral to the stimulus: although less diffuse than during wake, its association to the persistence of a minimal DMN frontal activation during points to a shallow sleep stage that only partially filters external input. In contrast, the complete disappearance both of the frontal component of the DMN and the abolition of the somato-sensory network indicates that deeper sleep stage are able to block disrupting incoming stimuli to the sleeping brain.