Robust and practical non-invasive estimation of local arterial wave speed and mean blood velocity waveforms.

Objective: Local arterial wave speed, a surrogate of vessel stiffness, can be estimated via the pressure-velocity (PU) and diameter-velocity (ln(D)U) loop methods. These assume negligible early-systolic reflected waves (RWes) and require measurement of cross-sectionally averaged velocity (U ), which is related to volumetric blood flow. However, RWes may not always be negligible and Doppler ultrasound typically provides maximum velocity waveforms or estimates of mean velocity subject to various errors (U ). This study investigates how these issues affect wave speed estimation and explores more robust methods for obtaining local wave speed and U .

Approach: Using aortic phase-contrast MRI (PCMRI, n  =  34) and a simulated virtual cohort (n  =  3325), we assessed errors in calculated wave speed caused by RWes and use of U rather than true U . By combining PU and ln(D)U loop wave speed values, (i) a corrected wave speed (ln(D)P), insensitive to RWes and velocity errors, was derived; and (ii) a novel method for estimating U from U was proposed (where U can be any scaled version of U ).

Main Results: Proof-of-principle was established via PCMRI data and in the ascending aorta, carotid, brachial and femoral arteries of the virtual cohort, with acceptably low wave speed and U errors obtained even when local pressure was estimated from diameter and mean/diastolic brachial pressures.

Significance: Given a locally measured diameter waveform and brachial cuff pressures, (i) the velocity- and RWes-independent ln(D)P method can be applied non-invasively and is likely more robust than ln(D)U and PU loop methods; and (ii) U can be estimated from routinely-acquired U .