Modern spectral analysis techniques for pulsed Doppler ultrasound velocity measurements

Fast Fourier transform (FFT), the maximum likelihood method (MLM), and autoregressive (AR) modeling with the Burg and modified covariance algorithms are compared for their suitability in deriving accurate blood flow velocity estimations from pulsed Doppler ultrasound measurements. Optimal parameter values are defined for each method, and the effects of signal-to-noise ratio (SNR) and signal bandwidth are investigated. It is shown that the MLM leads to the most robust velocity estimator.<
>