Dynamic spectral analysis of arterial Doppler blood flow signals using time-frequency representations

Abstract

Doppler signals from a disturbed arterial blood flow are very likely to be nonstationary for they usually have a time-varying frequency structure caused by a complicated flow pattern. Therefore conventional Fourier analysis based methods are unsuitable for this application. This paper introduces the application of joint time-frequency analysis (TFA) to estimate the frequency dynamics of arterial Doppler blood flow signals and to detect the presence of small stenoses of the artery. The short time Fourier transform (STFT), Wigner distribution (WD), and Choi-Williams distribution (CWD) were investigated. The results of synthetic signal analysis show that the STFT, WD and CWD follow the transient changes of frequency well, and the STFT and CWD also behave well for multicomponent signals. Doppler spectra were obtained by means of the FFT, STFT, WD, and CWD for different flow rates and positions along a simulated blood vessel for stenoses ranging from 2% to 25% cross sectional area reduction (CSAR) using a blood analogue. The results indicate that all of these methods are to some extent capable of detecting stenoses as small as 2% CSAR by identifying the existence of negative frequency components in the spectra, and the CWD appears at this stage to be superior to the STFT and WD.