A Comparison of Algorithms for the Detection of Stenotic Vessels

Abstract

In order to develop a senSItive detection scheme for minor stenoses, the received acoustic signal from regions within a stenosis and distal to a stenosis has been modeled and evaluated using experimental data. In comparison with a parabolic profile, analysis of the signal returned from vessel regions distal to the stenosis has shown that the correlated signal interval increases near the vessel wall and decreases in the center of the vessel.[l] With the goal of spatially mapping these changes to locate the source of the flow disturbance, two indicators of the correlated signal interval are presented and evaluated for known experimental flow conditions. The first parameter is the normalized likelihood function, evaluated at the maximum likelihood velocity, with coherent summation over a sequence of 8 ultrasonic pulses. The second parameter is the magnitude of the correlation at a lag of one pulse interval divided by the signal power. The experimental results show that the use of a parameter which coherently sums the signal correlation over a set of sequential pulses and lag values provides a sensitive indication of changes in a flow profile, and may be used to develop a spatial map of the flow conditions. This analysis differs from previous work in the transmission of a train of short, wideband pulses, and the resulting increased spatial resolution of the parameters.