An improved estimation and focusing scheme for vector velocity estimation

The full blood velocity vector must be estimated in medical ultrasound to give a correct depiction of the blood flow. This can be done by introducing a transversely oscillating pulse-echo ultrasound field, which makes the received signal influenced by a transverse motion. Such an approach was suggested in [1]. Here the conventional autocorrelation approach was used for estimating the transverse velocity and a compensation for the axial motion was necessary in the estimation procedure.

This paper introduces a new estimator for determining the two-dimensional velocity vector and a new dynamic beamforming method. A modified autocorrelation approach employing fourth order moments of the input data is used for velocity estimation. The new estimator calculates the axial and lateral velocity component independently of each other. The estimation is optimized for differences in axial and lateral modulation periods in the ultrasound field by using a lag different from one in the estimation process, and noise artifacts are reduced by using averaging of RF samples. Furthermore, compensation for the axial velocity can be introduced, and the velocity estimation is done at a fixed depth in tissue to reduce spatial velocity dispersion. Examples of different velocity vector conditions are shown using the Field II simulation program. A relative accuracy of 10.1 % is obtained for the lateral velocity estimates for a parabolic velocity profile for a flow perpendicular to the ultrasound beam and a signal-to-noise ratio of 20 dB using 20 pulse-echo lines per estimate. Performing the estimation on measured data from a sponge for a plug flow shows that the new estimator reduces the bias from -18 % to -8 %. The overall standard deviation averaged over all angles is reduced from 29.5 % to 10 %.