SVD and signal processing: algorithms, applications and architectures
SVD and signal processing: algorithms, applications and architectures
| Hi-index | 0.00 |
We have recently validated the use of a second-order Volterra (SoVF) model for separating the linear and quadratic components of the beamformed radio frequency (RF) data in pulseecho ultrasonic imaging. The quadratic component captures the second order nonlinearities in the echo data. Grayscale images from the quadratic component, referred to as quadratic B-mode (QB-mode) images, typically have higher contrast and increased dynamic range than the standard B-mode images (without loss in spatial resolution). An SVD-based algorithm for finding the coefficients of the SoVF based on a linear plus quadratic prediction model of the RF data was developed and validated in a variety of imaging situations, including in vivo. This approach, while successful, may produce images with ripple artifacts due to the truncation of the kernel. Adaptive SoVF implementations based on both LMS and RLS methods have been developed and are shown to produce artifact-free QB-mode images using relatively small kernels, i.e. well suited for real-time implementation. In this paper, we show results of imaging microbubble ultrasound contrast agents (UCAs) to demonstrate the increased sensitivity and specificity of QB-mode imaging compared to conventional methods.