Sommer F G, Stern R A, Howes P J, Young H
Med Phys. 1987 Jul-Aug;14(4):627-32. doi: 10.1118/1.596029.
Ultrasonic waveforms backscattered from tissue simulating phantoms and from normal and cirrhotic human livers in vivo were digitized to a standard dynamic range prior to envelope detection and determination of envelope amplitude distributions. For 11 individual narrow-band Gaussian-shaped filters of -6 dB bandwidth 200 kHz, and of center frequencies from 2 to 4 MHz, envelope amplitude distributions were plotted and mean values of the values distributions computed. Analysis of data was performed for data from a phantom containing only relatively small graphite scatters (less than 170 mu), and a similar phantom to which glass spheres 0.5 mm in diameter had been added homogeneously. For lower center frequency narrow-band filters, significantly more high-amplitude occurrences were observed for data from the phantom to which glass spheres had been added. Higher center frequency narrow-band filters gave significantly more high-amplitude occurrences for the phantom containing only small scatters. Similar data analysis was performed for in vivo human liver data from ten normal subjects and five patients with known cirrhosis of the liver. For the cirrhotic and normal livers, data analysis using narrow-band filters of relatively low center frequency resulted in more high- amplitude occurrences for cirrhotic, compared to normal liver; the converse was true for narrow-band filtration at relatively high center frequencies. Determination of mean amplitude following narrow-band filtration with a filter centered at 3.4 MHz was found to be quite repeatable for the normal and cirrhotic liver data; analysis of variance showed the measurement was 94.1% a function of the subject examined, and 5.9% related to the data acquisition session.(ABSTRACT TRUNCATED AT 250 WORDS)