Experimental Evaluation of Quantitative Diagnosis Technique for Hepatic Fibrosis Using Ultrasonic Phantom

Abstract

Since clinical diagnosis using ultrasonic B-mode images depends on the skill of the doctor, the realization of a quantitative diagnosis method using an ultrasound echo signal is highly required. We have been investigating a quantitative diagnosis technique, mainly for hepatic disease. In this paper, we present the basic experimental evaluation results on the accuracy of the proposed quantitative diagnosis technique for hepatic fibrosis by using a simple ultrasonic phantom. As a region of interest crossed on the boundary between two scatterer areas with different densities in a phantom, we can simulate the change of the echo amplitude distribution from normal tissue to fibrotic tissue in liver disease. The probability density function is well approximated by our fibrosis distribution model that is a mixture of normal and fibrotic tissue. The fibrosis parameters of the amplitude distribution model can be estimated relatively well at a mixture rate from 0.2 to 0.6. In the inversion processing, the standard deviation of the estimated fibrosis results at mixture ratios of less than 0.2 and larger than 0.6 are relatively large. Although the probability density is not large at high amplitude, the estimated variance ratio and mixture rate of the model are strongly affected by higher amplitude data.