Ultrasound-based Velocity Vector Imaging in the Carotid Bifurcation: Repeatability and an In Vivo Comparison With 4-D Flow MRI

Abstract

Objective

Ultrasound-based velocity vector imaging (US-VVI) is a promising technique to gain insight into complex blood flow patterns that play an important role in atherosclerosis. However, in vivo validation of the 2-D velocity vector fields in the carotid bifurcation, using an adaptive velocity compounding method, is lacking. Its performance was validated in vivo against 4-D flow magnetic resonance imaging (MRI). Furthermore, the repeatability of US-VVI was determined.

Methods

High frame rate US-VVI, which was repeated three times, and 4-D flow MRI data were acquired of the carotid bifurcation of 20 healthy volunteers. A semiautomatic registration of all US-VVI (n = 60) and 4-D flow MRI data was performed. The 2-D velocity vector fields were compared using cosine similarity and the root-mean-square error of the velocity magnitude. Temporal velocity profiles from the common carotid artery and internal carotid artery were compared. The interobserver and intraobserver agreement of US-VVI was determined for peak systolic velocities and end-diastolic velocities.

Results

The registration was successful in 83% of cases. The 2-D velocity vector fields matched well between modalities, which is supported by high cosine similarities and low root-mean-square error of the velocity magnitudes. Temporal profiles showed high resemblance, with similarity indices of 0.87 and 0.80, and mean peak systolic velocity differences of 0.91 and 7.9 cm/s in the common carotid artery and internal carotid artery, respectively. Good repeatability of US-VVI was shown with a highest bias and standard deviation of 1.7 and 11.7 cm/s, respectively.

Conclusion

Good agreements were found of both vector angles and velocity magnitudes between US-VVI and 4-D flow MRI. Given the high spatiotemporal resolution, US-VVI enables the capture of small recirculating regions of short duration that are missed by 4-D flow MRI.