Non-rayleigh first-order statistics of ultrasonic backscatter from normal myocardium

Abstract

Historically, it has been assumed that the first-order envelope statistics of ultrasonic backscatter from myocardium are best described by the Rayleigh probability distribution function (PDF); however, few studies have tested this assumption. This study compares the ability of five PDFs, including the Rayleigh, to describe high-frequency (5–15 MHz) ultrasonic backscatter from in vivo canine myocardium and in vitro human myocardium. The PDFs used have all been previously proposed as descriptions of backscatter from parenchymal tissue or other random media. In this study goodness-of-fit testing of backscatter recorded at several frequencies and insonification angles shows that only in a minority of cases is the envelope Rayleigh-distributed. In most cases, the K distribution provides the best fit. In addition, computer simulation shows that sparse Poisson scattering can produce K-distributed backscatter. Scattering of this type may contribute to the observed non-Rayleigh character of myocardial backscatter. We conclude that the Rayleigh distribution is not an adequate description of backscatter from myocardium. Non-Rayleigh distributions may offer an additional approach to myocardial tissue characterization.