Errors in phase velocity estimation owing to the method used for shear wave waveform phase extraction

Abstract

In ultrasound elastography, the shear wave phase velocity, c s , is often used to calculate the stiffness, which can indicate whether a biological tissue has an abnormality. The shear wave phase velocity can also be used to compute shear wave dispersion, which can be related to viscosity or guided wave effects. The most common methods used for extracting the phase from waveform vibration to estimate c s are the Kalman filter (KF), the fast Fourier Transform (FFT), and the Levenberg-Marquardt Algorithm (LMA). Currently, there are no studies in the literature comparing the performance of these methods; hence, this study aims to show the error in estimating the phase velocity obtained with the mentioned methods and to compare them through computational simulations and experiments. In the computational simulations, the absolute value of mean and the standard deviation of relative error of c s were evaluated as a function of signal-to-noise ratio, number of cycles of the shear wave, pulse repetition frequency, and initial phase. The experimental setup used an ultrasonic benchtop system, with a 2.104-MHz frequency transducer, T V , to generate acoustic radiation force at a plastic sphere and consequently induce shear wave propagation in a 7% gelatin phantom. Another ultrasonic benchtop system, with a 4.89-MHz frequency transducer, was used on pulse-echo mode to monitor the shear wave propagation at different positions within the phantom. The computational simulation results demonstrated that LMA and KF demonstrated the best performance in extracting the phase employed to estimate the shear wave phase velocity. In some situations, the c s values obtained by means of KF, FFT, and LMA methods in the experimental results presented a statistical difference, but the highest differences between the mean of c s was smaller than 0.03 m s −1 .