Parameter estimation for nonlinear frequency dependent attenuation in soft tissue.

Abstract

Most current attenuation estimation methods are based on an assumption that tissue attenuation is linearly frequency dependent. For tissues exhibiting nonlinear attenuation, such methods could produce biased results. This paper describes a new time-domain, amplitude method for estimating two parameters, beta and n, in nonlinear attenuation. In this method, echo signal of each A-line is first filtered in parallel by a bank of narrow-bandpass filters. Attenuation values at filter center frequencies can be calculated from the envelope peaks of each filtered narrowband signal, and beta and n are then estimated from these attenuation values. A quantitative study of the estimation uncertainty indicates that there is a significant increase in the variance of beta estimate when the attenuation model is changed from linear to nonlinear. The analysis yields insights into the tradeoff between estimation accuracy and estimation precision when tissue attenuation is only slightly deviated from linearity.