A new eigen-based flow estimator using the Matrix Pencil approach

Abstract

In color flow imaging, velocity estimation is often performed by first filtering out the clutter and then using a correlation-based estimator to find the mean flow velocity. As an alternative to this approach, a new eigen-based flow estimator called Matrix Pencil is presented in this paper. Matrix Pencil is a principal spectral component estimator that can be directly applied to the Doppler signal without prior clutter filtering, and it works by treating flow estimation as a generalized eigenvalue problem. To assess its performance, Matrix Pencil was used to perform velocity estimation on a set of raw color flow data obtained from a steady flow phantom. Our results showed that the Matrix Pencil estimates were close in location with the principal Fourier coefficients of the Doppler power spectrum. The color flow images obtained using Matrix Pencil also showed a similar flow gradient as that of the theoretical profile. As a comparison, the Matrix Pencil estimates were evaluated against the post-filter estimates obtained from an autocorrelator. It was found that the two approaches gave velocity estimates that were comparable to each other. © 2005 IEEE.