Aims: Pulse wave velocity (PWV) requires the length traveled between two aortic locations and the corresponding transit-time. This study aimed to: 1) design a new wavelet-based approach to estimate transit-time and aortic arch (aoPWV_W) from MRI ascending (AA) and descending (AD) aorta flow curves, and 2) compare its performances in terms of associations with: age, tonometric carotid-femoral PWV (cfPWV), and local AA Bramwell-Hill PWV (aaPWV_BH), against the time-domain aortic PWV (aoPWV_T), which is considered as the most relevant MRI approach.

Methods: We studied 74 healthy volunteers, (43±15years) who underwent MRI including: 1) multi-plane cine acquisitions to estimate aortic arch length and AA distensibility, converted into aaPWV_BH, and 2) through-plane velocity acquisitions which were automatically segmented to extract AA and AD flow curves, whose systolic-upslope was used to estimate transit-time using cross-correlation and the newly designed wavelet approach and subsequently to estimate aoPWV_T and aoPWV_W. First, cross-spectrum of AA and AD flow curves was calculated using wavelets which are more robust to insufficient temporal sampling of systolic-upslope than Fourier. Then the group delay was calculated by averaging the amplitude-weighed phases of systolic-upslope.

Results: Although strongly related (r=0.82, slope=1.15) associations with age, cfPWV and aaPWV_BH were stronger for the wavelet-based approach than aoPWV_T (aoPWV_W/aoPWV_T: r=0.76/0.63 with age; r=0.60/0.57 with cfPWV; r=0.56/0.49 with aaPWV_BH, p<0.001).

Conclusion: The wavelet-based arch PWV provides stronger associations with age and reference PWV than the previous time-domain estimate. These results suggest that considering supplementary signal information, results in a more reliable estimate of AA to AD transit time.