The gating role of the thalamus to protect sleep: an F-MRI report.

Objective. Neuroimaging is an innovative technique to study brain functiong during sleep. Most data regard nuclear neuroimaging, and only recently the potentials of functional Magnetic Resonance (fMRI) have been applied. Materials and methods. During an fMRI recording acquired on a 3 T scanner (MAGNETOM Allegra, Siemens, Erlangen, Germany) with contemporary electrical stimulation at the right median nerve at different frequencies (3 and 10 Hz), two subjects fell asleep during the 3 Hz session, and woke up when the frequency was shifted to 10 Hz. Functional data were analyzed using BrainVoyager and activated voxels identified with a GLM approach. Results. During electrical stimulation, subjects reported falling asleep, and were not responsive to commands. While asleep, BOLD signal increase was evident in thalamus, ascending-reticular-system and cerebellum, while BOLD signal decrease was observed over the bilateral occipital (Brodman areas 18 and 19), temporal (41) and posterior parietal cortex (7), with a minor area in the supplementary motor (6) and motor cortex (4). During wake, BOLD signal increase was recognized in the controlateral supplementary motor area (6), contralateral primary somatosensory cortex (7), bilateral secondary somatosensory cortex (5), bilateral insula and cerebellum. Discussion and conclusion. We report the first cases of fMRI of spontaneously sleeping healthy subjects during somato-sensory stimulation. A dissociation of BOLD signal activation/deactivation emerges: increased metabolism in thalamus and ascending-reticular-activating-system versus a BOLD decrease in heteromodal associative cortex (visual, auditory, parietal, prefrontal). This may be a visual rendering of the “thalamic gating hypothesis” of the sleeping brain. Reduced brain responsiveness during sleep depends on the disruption of signal transmission from the periphery to the cortex. As an integrating station of the sensitive pathways, the thalamus selects information to be projected or not to the pertinent cortices: in short, it protects the sleeping cortex. A limit of our data is the lack of EEG-coregistration. Sleep was recognized by the examiner and referred by subjects. Nonetheless, we can reliably believe the subjects fell asleep due to the activation-deactivation pattern congruent with previous data and the BOLD pattern mismatch between the wake/asleep scans.