CONTRAST AGENT DETECTION BASED ON AUTOREGRESSIVE SPECTRAL ESTIMATION

In this paper, we present a spectral autoregressive contrast detection method dedicated to the identification of ultrasound contrast agent in radiofrequency (RF) images, acquired with standard scanning devices used in echocardiography. This method is based on secondorder autoregressive modeling of the RF signal, and it relies on the presence of the second harmonic component. It is dedicated to processing single RF frames. Numerical simulations, as well as in vitro investigations show that the proposed approach enables to correctly detect the contrast agent, in particular at low concentrations of the agent. It is also shown that an associated technique of discontinuity adaptive smoothing enables to obtain a binary map of contrast agent-perfused areas.