Evaluation of flow estimation methods for 3D color Doppler imaging

In 3D ultrasound color Doppler imaging (CDI), similar to 2D CDI, 8-12 pulsetransmissions (ensembles, E) per each scanline are used for clutter rejectionand flow estimation, leading to a low volume acquisition rate. This rate couldbe improved by using a small number of ensembles (e.g., E=4). However, theimpact of the use of a small ensemble size on clutter rejection and flowestimation must be investigated. In this paper, we have evaluated three flowestimation methods: autoregression (AR), eigendecomposition (ED), andautocorrelation (AC) for a small size ensemble (E=4). To compare the performanceof three methods, raw radio-frequency (RF) data were acquired from the Dopplerflow phantom (RMI-1425A, Gammex Inc., USA) by using a commercial ultrasoundmachine (Accuvix V10, Medison Corp., Korea) equipped with a research package.From the phantom experiment, the AR estimator exhibited a significantimprovement in area under receive operating characteristic curves compared to ACand ED methods (i.e., 0.97 ± 0.02 vs. 0.86 ± 0.07 and 0.83± 0.05, respectively). There is no statistically significant differencebetween the AC and ED methods (p>0.24). These results indicate that the ARestimator would be suitable for flow estimation in 3D CDI with a low ensemblesize.