Efficacy of Pearson distributions for characterization of gray levels in clinical ultrasound kidney images

Abstract

A major issue in characterization of clinical Ultrasound B-mode images has been to identify a few probability distributions which can broadly classify images of human organs. This paper precisely addresses the concern by resorting to statistical characterization of gray levels in 100 clinical B-mode ultrasound kidney images. A detailed investigation of central and peripheral kidney regions is carried out. Pearson family of distributions is employed as a means to model the images and to provide analytical expression of the suitable pdf in terms of the parameter $$\kappa $$ . In majority of cases, it is noted that type I Pearson distribution (corresponding to negative $$\kappa $$ ) yields the best fit. For testing the efficacy of Pearson family of distributions in relation to commonly used probability distributions viz. Nakagami, Inverse Gaussian, Gamma, and Weibull, performance measures based on Jensen–Shannon divergence and Kolmogorov–Smirnov test statistics are used. The statistical analysis reveals that Pearson family of distributions outperforms which is evident through cumulative ranking of probability distributions considered in the study. This study attempts to identify the Pearson types which can universally characterize the clinical US B-mode images of different human organs.