Variance Reduction for Ultrasonic Attenuation Measurements from Backscatter in Biological Tissue

Absfracf-Various averaging techniques can be used to reduce the variability of frequency-dependent attenuation estimates that are based on the power spectral density of acoustic backscatter signals from biological tissue. Three variance reduction methods are compared in terms of a consistently defined confidence interval. It is shown that the attenuation estimate based on averaged spectra from two C-planes is superior to the averaged attenuation estimate based on the pairs of spectra from individual A-lines. The superior technique yields results of equivalent accuracy with approximately 40 percent less data. This method is also shown to be asymptotically optimal in the sense of maximum likelihood estimation.