Two-dimensional blood velocity estimation with ultrasound: speckle tracking versus crossed-beam vector Doppler based on flow simulations in a carotid bifurcation model

Detailed imaging of complex blood flow may improve early diagnosis of cardiovascular disease. In clinical practice, non-invasive flow imaging has been limited to onedimensional Doppler techniques. Searching for multi-dimensional estimators, research has given attention to speckle tracking (ST) and vector Doppler (VD). However, these techniques have yet to be validated for complex flow patterns as may arise in diseased arteries. In this work, the properties of ST and crossed-beam VD are compared with a ground truth for clinically relevant flow using an ultrasonic simulation environment coupled with the output from computational fluid dynamics (CFD).The statistical properties (N=80) of ST and VD were firstevaluated for stationary flow in a tube for varying vessel positions and angles, and for varying noise levels. The parameter study demonstrated VD to be a more robust axial velocity estimator, while similar results were overall obtained for the lateral velocity component. As an example, the relative standard deviation was 15% and 8% for ST compared to 3% and 10% for VD, for the axial and lateral velocity component respectively.Further, performance was evaluated for pulsatile flow conditions in a stenosed carotid bifurcation model. A linear regression analysis showed that both methods overall had a good agreement to the CFD reference, however VD suffered from more spurious artifacts and was severely hampered by aliasing in parts of the cardiac cycle. ST was less accurate in estimating the axial component, but prevailed in estimating velocities well beyond the Nyquist range. Based on our simulations, both methods may be used to image complex flow behavior in the carotid bifurcation, however, considering also scanning limitations of VD, ST may provide a more consistent and practical approach. Future work will entail in-vitro and in-vivo validation of these results.Detailed imaging of complex blood flow may improve early diagnosis of cardiovascular disease. In clinical practice, non-invasive flow imaging has been limited to onedimensional Doppler techniques. Searching for multi-dimensional estimators, research has given attention to speckle tracking (ST) and vector Doppler (VD). However, these techniques have yet to be validated for complex flow patterns as may arise in diseased arteries. In this work, the properties of ST and crossed-beam VD are compared with a ground truth for clinically relevant flow using an ultrasonic simulation environment coupled with the output from computational fluid dynamics (CFD).The statistical properties (N=80) of ST and VD were firstevaluated for stationary flow in a tube for varying vessel positions and angles, and for varying noise levels. The parameter study demonstrated VD to be a more robust axial velocity estimator, while similar results were overall obtained for the lateral velocity component. As an example, the relative standard deviation was 15% and 8% for ST compared to 3% and 10% for VD, for the axial and lateral velocity component respectively.Further, performance was evaluated for pulsatile flow conditions in a stenosed carotid bifurcation model. A linear regression analysis showed that both methods overall had a good agreement to the CFD reference, however VD suffered from more spurious artifacts and was severely hampered by aliasing in parts of the cardiac cycle. ST was less accurate in estimating the axial component, but prevailed in estimating velocities well beyond the Nyquist range. Based on our simulations, both methods may be used to image complex flow behavior in the carotid bifurcation, however, considering also scanning limitations of VD, ST may provide a more consistent and practical approach. Future work will entail in-vitro and in-vivo validation of these results.A