A new approach for determining maximum frequency in clinical Doppler ultrasound spectral estimates

Accurate estimation of maximum frequency from a clinical Doppler ultrasound power spectrum is an important tool in the diagnosis of arterial stenosis severity. In this study, the sensitivity of signal to noise ratio (SNR) of four established maximum frequency estimators (percentile method, modified threshold crossing method, hybrid method, and geometric method) are compared with a new algorithm (signal-noise slope intersection method). Steady flow signals from a clinical Doppler ultrasound machine were used to calculate power spectrums via an FFT from which the maximum frequencies were estimated, Variations in the SNR and intrinsic spectral broadening were obtained by adjusting the Doppler gain and receive aperture size. In general, the geometric method had better repeatability than the other three established methods except at low SNRs. In comparison, the proposed signal-noise slope intersection method was more consistent and illustrated excellent repeatability over the experimental range of SNRs.