Using backscattered ultrasound signals to characterize normal human liver tissue periodicity with the spectral autocorrelation method

Nowadays the spectral analysis of the radio-frequency (RF) signal from biological tissues using ultrasound (US) is a promising research topic regarding the acquisition of quantitative diagnostic information. The mean scatterer spacing (MSS) has been studied in the characterization of medium periodicity. In order to estimate MSS, it has been proposed the spectral autocorrelation (SAC) method (Varghese and Donohue, 1993). This technique basically generates a matrix in the frequency domain from the ultrasound signals, where the periodicity is related to off-diagonal peaks. The present work aims to use SAC to characterize in vitro normal human liver periodicity using backscattered 20-MHz ultrasound signals. Five human hepatic samples (about 7-mm thick) were studied. They have been previously classified as normal after biopsy. A total of 300 ultrasound backscattered signals were collected from each specimen, and then processed with the SAC method. The MSS is estimated from the frequency corresponding to the spectral peak. Histograms of the estimates were constructed to analyze distribution patterns. The MSS estimates are compatible with literature (mean MSS = 1,02 ± 0,08 mm, considering the five samples together). The histograms show several specific modes. Comparing SAC results with other spectral analysis methods from literature, although the limited number of specimens, it can be seen that there is a good agreement in estimates, being the MSS around 1 mm (which is also compatible with the medical literature). Nevertheless, the several important modes observed in the histograms may represent various periodicities inside the tissue. The next step is to analyze fibrotic hepatic tissues, and to compare them with the results here presented, to see if it is possible to differentiate the normal and pathological conditions.