Investigation of the estimation accuracy of two-step block matching methods using envelope and RF signals for two-dimensional blood flow vector imaging

In this paper we investigate the estimation accuracy of two-step block matching methods, which perform a rough detection corresponding to pixel-level displacement in the first step and then a finer detection using a spatial interpolation in the second step. In the present study, four strategies were examined: (1) RF signals in both steps; (2) envelopes of RF signals in both steps; (3) envelopes in the first step and RF signals in the second step; (4) RF signals in both steps with an extended interpolation region. Their performances were evaluated by phantom and simulation experiments. In terms of the estimation accuracy and calculation time, strategy (3) was preferable for two-dimensional blood flow vector imaging. In in vivo measurements of the common carotid artery using this method, a blood flow velocity of approximately 650 mm s^-1 was observed at the center and vortex flow at the bifurcation was visualized.