J. Martí et al. (Eds.): IWDM 2010, LNCS 6136, pp. 452–458, 2010.
© Springer-Verlag Berlin Heidelberg 2010
An Anthropomorphic Software Breast Phantom for
Tomosynthesis Simulation: Power Spectrum Analysis of
Phantom Projections
Predrag R. Bakic
1
, Beverly Lau
2
, Ann-Katherine Carton
1
, Ingrid Reiser
2
,
Andrew D.A. Maidment
1
, and Robert M. Nishikawa
2
1
University of Pennsylvania, Dept. of Radiology, Philadelphia, PA 19104 ,
2
University of Chicago, Dept. of Radiology, Chicago, IL 60637
{Predrag.Bakic,Ann-Katherine.Carton,
Andrew.Maidment}@uphs.upenn.edu
{Beverly,IReiser,R-Nishikawa}@uchicago.edu
Abstract. We have performed spectral analysis of simulated tomosynthesis pro-
jections generated using an anthropomorphic software breast phantom. Twenty
phantoms were generated: ten 450 ml phantoms with 40% glandular fraction
and ten 1500 ml phantoms with 20% glandular fraction. Simulated mammo-
graphic compression and acquisition was performed using monoenergetic ray-
tracing. ROIs were extracted and the modulus-squared 2D FFT was applied to
each ROI to obtain periodograms. Radially-averaged periodograms were com-
pared between phantom and clinical images. We observed a good agreement
between the spectral power law exponents (β) calculated from phantom projec-
tions and clinical images.
Keywords: breast anthropomorphic phantom, digital breast tomosynthesis,
power law, power spectrum.
1 Introduction
Clinical validation of novel breast imaging systems is largely unfeasible today as it
requires long and expensive clinical trials. On the other hand, physical characteristics
of imaging systems such as the MTF, NNPS and NEQ, do not necessarily predict the
behavior of the human observer scrutinizing complex mammographic backgrounds
and do not take into account clinical processing or display. In an alternative approach,
a voxelized anthropomorphic software breast phantom has been developed for use in
pre-clinical validation of breast imaging modalities. The phantom realistically simu-
lates the spatial distribution of adipose and fibroglandular tissues with known ground
truth in simulated images. Projections of simulated tissue structures generate realistic
parenchymal pattern, called anatomical noise. The anatomical noise is known to affect
visibility of breast lesions. [1].
A frequently used descriptor of the anatomical noise is the power law exponent (β)
of the radially-averaged periodogram. A periodogram is the modulus-squared 2D dis-
crete Fourier transform of a region of interest (ROI). Burgess et al. [2] demonstrated