Comparison between cross-correlation and auto-correlation technique in color flow imaging

It has previously been reported that the RF signal cross-correlation technique gives a lower variance estimate of blood velocity than the conventional baseband auto-correlation technique. This can be explained by the axial averaging used in calculating the cross correlation, which also reduces the spatial resolution. If similar axial averaging is applied to the auto-correlation estimate, it is shown both by theory and experiments, that the two methods are almost identical. Estimation variance is calculated using a theoretical signal model, and results are verified by applying the different algorithms to digitized ultrasound RE-data. Optimum pulse- and receiver bandwidth, as a function of signal-to-noise ratio is discussed