Sleep Alters the Velocity of Physiological Brain Pulsations in Humans

The clearance of brain metabolites increases during sleep, in association with increased spectral power of the three main cerebrospinal fluid (CSF) flow drivers: cardiovascular, respiratory, and vasomotor brain pulsations. However, little is known about how the increased power of these pulsations affects the velocity and direction of fluid flow in the sleeping brain.

To address this knowledge gap, we mapped the CSF oscillatory flow velocity in relation to the changing physiological pulsations in the brains of 22 healthy volunteers during sleep and waking.

We used the ultrafast magnetic resonance imaging sequence known as magnetic resonance encephalography (MREG) for tracing the pulsatile movement of water molecules inside the cranium. First, we conducted a phantom validation study with optical flow analysis to confirm that MREG accurately tracks pulsatile water molecule flow in a porous tissue medium. Next, we obtained MREG recordings for mapping the three physiological pulsations without aliasing in the human brain across the awake and sleep states; we thereby quantified the brain-wide 3D velocity $\underset{V}{\to }$ vectors (i.e., the velocity v_s and 3D direction $\hat{v}$ ) of each pulsation band, using comprehensive dense optical flow analysis during EEG-verified sleep in comparison to the awake state. Finally, we assessed relationships among the spectral power of the physiological pulsations, their 3D velocity $\underset{V}{\to }$ , and slow-delta EEG power, which is known to depict the increased interstitial volume during sleep.

In our phantom study, dense optical flow analysis reliably detected water flow in tissue driven by external pulsations. In healthy volunteers, sleep increased flow velocities ( $\underset{V}{\to }$ ) of the pulsations by more than 20% in concert with elevations in respiratory pulsations and vasomotor waves, while the velocity of cardiovascular pulsations (v_s) declined by the same percentage. There was a significant anticorrelation between cardiac mean spectral power and slow delta EEG mean power, and a significant correlation between vasomotor mean spectral power and slow delta EEG mean power over the whole brain.

Phantom studies validated the optic flow analysis of fast MREG recordings. Sleep altered the 3D velocity dynamics of all neurofluidic brain pulsations in a manner consistent with increased interstitial space and greater fluid exchange, thus supporting the glymphatic model wherein physiological pulsations drive bulk flow during sleep.