Comparison of modeling accuracy of amplitude distribution models for ultrasonic tissue characterization of liver fibrosis

A quantitative diagnostic method for liver fibrosis using ultrasound echo signals is highly required. A probability density function (PDF) of echo envelope from a normal liver can be approximated by a Rayleigh distribution; however, the PDF of echo envelope from liver fibrosis deviates from the Rayleigh distribution. To evaluate tissue characteristics in the ultrasound B-mode image, several amplitude distribution models have been proposed. We proposed a multi-Rayleigh distribution model and evaluation method of liver fibrosis using the multi-Rayleigh model. In this study, we evaluated the modeling accuracy of the multi-Rayleigh model and other amplitude distribution models using the KL divergence. From the evaluated results for the 120 clinical data, it was found that the multi-Rayleigh model with three components is more suitable model than other amplitude distribution models for approximating the PDF of echo envelope from the liver fibrosis.