An Improved Mean Frequency Estimator for Ultrasonic Color Flow Imaging Using Second-Order Autoregressive Model

An improved second-order AR estimator is proposed for color flow imaging. It can give accurate estimation in the presence of very intensive clutter signal. Traditionally, the performance of the second-order AR estimator may be degraded by the clutter when the amplitude of the clutter is very large compared with the Doppler signal, because very large clutter will shift the Doppler pole toward low frequency. The improved second-order AR estimator which we propose in this paper can remedy this problem by first estimating the mean frequency of the clutter signal and then compensating the frequency down-shift of the Doppler signal using this mean frequency. Simulation results show that the improved AR estimator can give more accurate estimate of the mean frequency and variance of the blood signal than the conventional one, even if the amplitude of the clutter is very large compared with Doppler signal