Ultrasoundin Med. & Biol. Vol. 15, No. 3, pp. 263-272, 1989 0301-5629/89 $3.00 + .00 Printed in the U.S.A. © 1989 Pergamon Press plc OOriginal Contribution A REAL-TIME AUTOREGRESSIVE SPECTRUM ANALYZER FOR DOPPLER ULTRASOUND SIGNALS F. S. SCHLINDWEIN Universidade Federal do Rio de Janeiro, RJ, Brazil and D. H. EVANS Department of Medical Physics and Clinical Engineering, Leicester Royal Infirmary, Leicester LE1 5WW, Englandt (Received 13 June 1988; in final form 28 September 1988) AbstractmA system based on a digital signal processor and a microcomputer has been programmed to estimate the maximum entropy autoregressive (AR) power spectrum of ultrasonic Doppler shift signals and display the results in the form of a sonogram in real-time on a computer screen. The system, which is based on a TMS 320C25 digital signal processor chip, calculates spectra with 128 frequency components from 64 samples of the Doppler signal. The samples are collected at a programmable rate of up to 40.96 kHz, and the computation of each spectrum takes typically 3.2 ms. The feasibility of on-line AR spectral estimation makes this type of analysis an attractive alternative to the more conventional fast Fourier transform approach to the analysis of Doppler ultrasound signals. Key Words: Ultrasound, Ultrasonic Doppler shift, Doppler power spectrum, Autoregressive spectrum analysis, Fast Fourier transform analysis. 1. INTRODUCTION Doppler ultrasound, both pulsed and continuous wave, alone and in conjunction with ultrasonic imaging, is widely used as a noninvasive method for the assessment of blood flow. The Doppler shift fre- quency is proportional to the velocity of the blood within the sample volume, and since arterial blood flow is pulsatile, the Doppler shift signal has a spec- trum that is constantly varying with time. Under ideal conditions the Doppler power spectrum has a similar shape to a histogram of the blood velocities within the sample volume and thus real-time spectral analysis of the Doppler signal produces information concerning the evolution of the velocity distribution in the artery. The estimation of the power spectral density of a Doppler signal is normally performed by applying a fast Fourier transform (FFT) directly to the sampled signal and squaring the magnitudes of the output values. The signal is processed in individual frames of t" Address for correspondence. 263 N samples (usually 128 or 256), and is usually weighted, using one of the many window functions known to minimize spectral leakage, before the FFT is calculated. This technique of transforming the data directly is normally referred to as the periodogram approach. The most common way of displaying the evolution of the spectral information is the Doppler sonogram, in which the Doppler shift frequency is charted along the vertical axis, time along the hori- zontal axis, and the signal power (corresponding to a particular velocity and time) as a change of intensity on a screen or, for hard-copies, on paper (Fig. 1). The periodogram is computationally a very effi- cient method and produces fairly good results for the typical analysis regime used (128 frequency compo- nents every 5 or 10 ms), but it does have its limita- tions: The frequency resolution is the reciprocal of the time interval over which each frame is sampled, and the discrete Fourier transform implicitly assumes that the data is periodic and that an integer number of periods are used. Since this is generally not the case, the implicit periodicity causes a discontinuity that does not belong to the original signal, and the