Analysis of laser Doppler flux motion in man: comparison of autoregressive modelling and fast Fourier transformation.

Abstract

In order to investigate laser Doppler (LD) flux motion in healthy subjects and patients with peripheral arterial occlusive disease (PAOD), spectrum analysis of LD signals is needed. Autoregressive analysis (AR) is presented as an alternative method of power spectrum estimation. This procedure is compared to the commonly used fast Fourier transform algorithm (FFT) by describing the analytical power of both spectra in the analysis of flux motion waves. LD signals were recorded from the forefoot of 8 healthy volunteers and 11 patients with different degrees of PAOD. The flux, concentration of moving blood cells and velocity signal was digitized and stored for off-line analysis. Special software was designed to calculate AR and FFT spectra of the LD signals and to compare the suitability of both methods for the spectral analysis of LD recordings. Additionally, three-dimensional spectrum diagrams were calculated to demonstrate time-dependent flux changes during standardized provocation maneuvers. AR facilitates the determination of frequency and amplitude of flux motion waves as compared to the FFT. Low frequency-large amplitude waves (LF waves) were detected in both groups. High frequency-small amplitude waves (HF waves), which predominantly appear in severe ischemia, were observed in 7 of the 11 patients and in 2 of the 8 controls. The spectra revealed pulsatile waves in all healthy controls, but only in 1 of the 11 patients. AR modelling allows a reliable description of important flux motion components and has considerable advantages in spectral estimation of LD signals as compared to the FFT.