ABSTRACT A theoretical analysis was performed to describe statistical characteristics of calibrated spectral parameters used for ultrasonic tissue evaluation in the prostate and liver. The analysis assumes that radio-frequency (rf) echo signals exhibit Gaussian statistics. It derives the probability density function (pdf) of spectral parameters that are computed using sliding-window analysis techniques. The analysis relates the standard deviations of linear-regression spectral-parameter estimates to system and analysis parameters including bandwidth, center frequency, and the length of the sliding analysis window. The analysis also derives the pdf for mid-band fit parameter images. Theoretical results are found to agree well with clinical data from homogeneous segments in liver and prostate. The results offer a basis for evaluating spectral-estimator precision and for conducting future studies of lesion detectability based on spectral features.
1997 IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 44:935-942
© 1997 by The Institute of Electrical and Electronics Engineers, Inc. All rights reserved.