We studied the influence of changes in gain settings, log compression,
persistence, preprocessing, and postprocessing on image density in th
e fetal liver model. Each parameter was studied while the others were
held constant. The image density was objectively measured by electroop
tical transmission densitometry using a transparent film output. Neith
er the persistence nor the preprocessing levels significantly changed
image density. Postprocessing of sonographic images produced significa
nt differences (p < 0.02) in mean image density of most of the various
postprocessing curves and is a serious confounder of tissue echogenic
ity assessment. The data reconfirmed that there is a linear relationsh
ip (r = -0.94 to -0.997) between image density and gain setting. Howev
er, each log compression setting significantly changed (p < 0.0000001)
this relationship, obviating possible image density calibration. Our
data suggest that manipulation of image parameters by computed sonogra
phic technology obviates accurate echogenicity assessment.