Comparison of different attenuation compensation methods in the estimation of the ultrasound backscatter coefficient

Abstract

This study applied three methods to compensate for intervening tissue attenuation in the estimation of the backscatter coefficient of a tissue-mimicking phantom with spatial variations of attenuation. The first method, considered as the gold standard, made use of attenuation estimates measured using an independent substitution technique. The second method estimated first the local attenuation using a Constrained-Spectral Log Difference (CLD) technique, and then used these values to quantify the total intervening tissue attenuation. The third method consisted on a regularized, dynamic programming technique that simultaneously computes the backscatter coefficient and the total attenuation. Two variants of the latter method were investigated, one with search ranges centered around the expected attenuation and backscatter values of the phantoms, and a second one with search reaches expanded to include attenuation and backscatter values expected in different breast tissues. Our results indicate that, although the CLD method provided more precise estimates of the local attenuation, the DP method with search ranges centered around the expected phantom values provided more accurate attenuation compensation and backscatter coefficient estimation. When increasing the search ranges, the accuracy and precision of DP was importantly reduced.