Fig. 3

CSF in the perivascular space is pumped by arterial wall motions. a Representative traces of ECG (red curve), root-mean-square velocity v_rms (blue curve), and respiration (green curve). b The v_rms conditionally averaged over the cardiac and respiratory cycles³⁸. The protruding plots show v_rms averaged over the cardiac (left) or respiratory (right) cycle alone. c (inset) An example illustrating how Δt, the delay time between adjacent peaks in the v_rms and ECG/respiration, is calculated. Average probability density functions of Δt for ECG (red) and respiration (green), with shaded regions indicating standard error of the mean (SEM). While there is no clear trend for the occurrence of a peak in v_rms after a respiration peak, the v_rms peaks are most likely to occur soon after an ECG peak. d Synchronized measurements of the ECG (red curve) and artery diameter (black curve), obtained from transversal line scans of the pial arteries; the blue dashed line is a rolling average of the artery diameter. e The normalized average change in the artery diameter (i.e., the shape of the arterial wall traveling wave) averaged over the cardiac cycle. Mean ± SEM, n = 7 mice. f The arterial wall velocity (black curve) obtained by calculating the average derivative of the change in artery diameter over the cardiac cycle, and Δv_rms (blue curve) obtained by calculating the difference in v_rms from its mean over the cardiac cycle. Mean ± SEM, n = 7 mice. The maximum arterial wall velocity (21.2 ± 3.7 µm s⁻¹, n = 7) agrees well with the peak Δv_rms value (9.9 ± 3.8 µm s⁻¹, n = 8) and occurs 34.5 ± 7.4 ms after the ECG peak, indicating that perivascular pumping is most likely the principal driving mechanism