Ultrasound attenuation measurements of the liver in vivo using a commercial sector scanner.

Attenuation measurements of various tissue mimicking phantoms and three different groups of patients were obtained using a modified commercial sector scanner. Estimates of attenuation were made using the spectral shift method with mean frequencies at different depths of a region of interest being obtained by both zero crossing and fast Fourier transform techniques. The accuracy and precision of both techniques was compared in phantoms and it was found that the FFT technique yielded less day-to-day variation (SD=3 percent) than the zero crossing technique (5 percent). For larger regions of interest, the range of variation in both techniques was more similar. Day-to-day variation in livers of normal patients was much larger than that seen in phantoms (10 to 15 percent) suggesting that in vivo measurements may be less precise due to actual daily changes in patients' livers. Attenuation estimates of phantoms were high by approximately 0.16 dB/MHz/cm compared to values obtained by transmission techniques. The attenuation values of livers in a group of 31 normal patients ranged from 0.214 dB/cm/MHz to 0.849 dB/cm/MHz with a mean of 0.627 +/- 0.126 dB/cm/MHz for the zero crossing technique while the mean value using the FFT technique was 0.86 +/- 0.168 dB/cm/MHz. A group of 26 Gauchers disease patients also showed wide variation with a mean attenuation value of 0.768 +/- 0.21 dB/cm/MHz using the FFT technique. This was significantly different than that of the normal group (p less than .05). Also, a group of 22 chronic B hepatitis patients was examined, having a mean attenuation value of 0.823 +/- 0.21 dB/cm/MHz, not significantly different from those of normal patients. Highly significant differences were found between the three groups when the power spectrum bandwidths of signals received were compared. These differences may be due to differences in the dependence of attenuation as a function of frequency between the groups and may represent a useful tissue characterization parameter.