Finding the peak velocity in a flow from its doppler spectrum

The signal backscattered by blood cells crossing a sample volume produces a Doppler power spectrum determined by the scatterers' velocity distribution. Because of intrinsic spectral broadening, the peak Doppler frequency observed does not correspond to the peak velocity in the flow. Several methods have been proposed for estimating the maximum velocity component-an important clinical parameter-but these methods are approximate, based on heuristic thresholds that can be inaccurate and strongly affected by noise. Reported here is a method of modeling the Doppler power spectrum of a flow, and from that model, determining what Doppler frequency on the descending slope of the power spectrum corresponds to the peak velocity in the insonated flow. It is shown that, for a fully insonated flow with a parabolic velocity distribution, the peak velocity corresponds to the Doppler frequency at the half-power point on that slope. The method is demonstrated to be robust with regard to the effects of noise and valid for a wide range of acquisition parameters. Experimental maximum velocity measurements on steady flows with rates between 100 and 300 mL/min (peak velocity range 6.6 cm/s to 19.9 cm/s) show a mean bias error that is smaller than 1%.