Affiliations: Digital Imaging Research Division, Department of
Radiology, Duke University Medical Center, Durham, NC, USA | Department of Biomedical Engineering, Duke University,
Durham, NC, USA
Note: [] Corresponding author: Alan H. Baydush, PhD, Box 2623, Duke
University Medical Center, Durham, NC 27710, USA. Tel.: +1 919 684 2691; Fax:
+1 919 684 3934; E-mail: alan.baydush@duke.edu
Abstract: {\it Purpose:} The effect of Bayesian image estimation (BIE), which
reduces image scatter and improves contrast-to-noise ratios (CNR), on reduced
dose digital mammography is studied. {\it Methods/Materials:} Standard dose digital mammographic images
of the ACR breast phantom were acquired with and without a grid. Additional
images were acquired without the grid at reduced exposures and were processed
by BIE. CNR values were computed for visible masses before and after processing
for each exposure and compared to using a grid. {\it Results:} With no change in exposure, BIE applied to the image
acquired without the grid improves CNR from 209 to 520% over that provided by
the grid. While these results were not as good at reduced exposures, BIE
processing still improved CNR from 236 to 465% over a grid. {\it Conclusions:} BIE processing of reduced exposure level images
offers increased CNR and subjective visualization compared to images acquired
with a grid at standard exposure.
Keywords: scatter, image processing, statistical estimation, Bayesian, digital mammography
