Feature analysis of Doppler ultrasound signals obtained from mammalian arteries

In this research, we have derived a closed-form solution for the axial and radial velocities of a Newtonian incompressible fluid flowing under a pulsatile pressure gradient through nonuniform, cylindrical, rigid and elastic tubes. The nonuniformity was achieved by choosing an obstruction having the shape of a {dollar}beta{dollar}-function on the inner surface of the tube. The theory thus developed was applied to mathematically model the flow of blood through arteries with fat and cholesterol deposits on their inner surface. The advantage of using a {dollar}beta{dollar}-function to model the internal obstruction, was that it could be skewed to either side so as to more realistically approximate the various shapes and sizes of plaque deposits in diseased arteries.; From the mathematical model we simulated a Doppler signal having the same basic characteristics of amplitude and frequency modulation as an experimental Doppler ultrasound signal that is obtained from diagnostic ultrasound equipment used in hospitals. We used a Diasonics ultrasound scanner to obtain the experimental Doppler signals from the carotid arteries of humans, as well as from a laboratory model that utilized polyurethane and silicone tubes having varying degrees of elasticity.; Several signal processing methods were applied to both the simulated and experimental Doppler signals to study the effect of atherosclerosis on hemodynamics. These methods included the exponential time-frequency distribution and the wavelet transform. In addition, we also used the cepstrum coefficients, and the energy and autocorrelation functions. The goal was to determine the degree (% reduction of lumen diameter), position (anterior, posterior, medial or lateral), and the shape of atherosclerotic plaque in human carotid arteries. Using our fluid flow model, it was possible to change any of its parameters (for example, the shape or size of the obstruction, or the elasticity of the vessel wall) and note the corresponding effect on any feature extracted from the simulated Doppler signal. Whilst other researchers have focused primarily on determining the presence of the plaque and on the degree of reduction of lumen diameter, we have also aimed at determining the plaque shape and its position.; The results of our simulations using the mathematical model are presented. We monitored the changes in the velocity profiles of the fluid as a function of time, as it flowed through the obstruction. We also noted that the propagation velocity of the pressure pulse wave in an elastic tube decreased, whereas the attenuation per wavelength increased, as the wave traveled from the proximal end of the obstruction to the location in the region of the obstruction where the tube diameter was minimum. Cepstral analysis was useful in determining the degree of blockade by measuring the position, in feature space, of the observed feature with respect to that of clusters of features of normal and stenotic arteries of known degrees of obstruction. Using the autocorrelation function we had limited success in determining the location of the plaque. However, it proved to be very useful in determining the plaque position. The exponential distribution was used to monitor the changes in the frequency content of the Doppler signal during a complete pressure cycle. By measuring the ratio of energies of the wavelet coefficients corresponding to various subbands, the wavelet transform proved effective in determining the approximate shape of the obstruction. (Abstract shortened by UMI.)...