Validation and Optimization of Digital Breast Tomosynthesis
Reconstruction using an Anthropomorphic Software Breast Phantom
Predrag R. Bakic, Susan Ng*, Peter Ringer*, Ann-Katherine Carton,
Emily F. Conant, and Andrew D.A. Maidment
University of Pennsylvania, Department of Radiology, 3400 Spruce St., Philadelphia PA 19104
*Real-Time Tomography, LLC, 1709 Balsam Lane, Villanova, PA 19085.
[Predrag.Bakic | Ann-Katherine.Carton | Emily.Conant| Andrew.Maidment] @uphs.upenn.edu
*[Susan.Ng | Peter.Ringer] @realtimetomography.com
ABSTRACT
A digital breast tomosynthesis (DBT) reconstruction algorithm has been optimized using an anthropomorphic software
breast phantom. The algorithm was optimized in terms of preserving the x-ray attenuation coefficients of the simulated
tissues. The appearance of the reconstructed images is controlled in the algorithm using three input parameters related
to the reconstruction filter. We varied the input parameters to maximally preserve the attenuation information. The
primary interest was to identify and to distinguish between adipose and non-adipose (dense) tissues. To that end, a
software voxel phantom was used which included two distinct attenuation values of simulated breast tissues. The
phantom allows for great flexibility in simulating breasts of various size, glandularity, and internal composition.
Distinguishing between fatty and dense tissues was treated as a binary decision task quantified using ROC analysis. We
defined the reconstruction geometry to enable voxel-to-voxel comparison between the original and reconstructed
volumes. Separate histograms of the reconstructed pixels corresponding to simulated adipose and non-adipose tissues
were computed. ROC curves were generated by varying the reconstructed intensity threshold; pixels above the
threshold were classified as dense tissue. The input parameter space was searched to maximize the area under the ROC
curve. The reconstructed phantom images optimized in this manner better preserve the tissue x-ray attenuation
properties; concordant results are seen in clinical images. Use of the software phantom was successful and practical in
this task-based optimization, providing ground truth information about the simulated tissues and providing flexibility in
defining anatomical properties.
Keywords: Mammography, digital breast tomosynthesis, anthropomorphic phantom, x-ray image simulation,
tomographic reconstruction, optimization of imaging systems.
1. INTRODUCTION
Digital breast tomosynthesis (DBT)
1
is undergoing final system development and initial clinical trials.
2, 3
Optimization
of DBT is typically based upon the use of physical measures, or subjective comparison of clinical images. While
clinical trials represent the preferred validation approach they pose a practical limitation; it is not feasible to conduct
clinical trials for a large number of system configuration combinations. We have developed a preclinical optimization
method based upon the analysis of simulated images of an anthropomorphic breast software phantom. The optimization
method is well suited for quantitative assessment as the phantom provides ground truth about the spatial distribution of
simulated tissue and tissue properties.
The goals of this research are to validate and optimize DBT reconstruction methods. Our initial effort was to
reconstruct images which best portray linear x-ray attenuation coefficients of the breast tissue for the task of estimating
breast density, an image-based biomarker of breast cancer risk.
4
This work is an extension of our previous analysis of
dense tissue regions extracted from clinical tomosynthesis images.
5-7
In our previous research, we calculated the spatial
correlation between regions of dense tissue segmented from the orthogonal DBT projection image and the central
reconstructed image.
6
While changing the DC component of the DBT reconstruction filter frequency response, we
searched for the best spatial matching (estimated using the Jaccard coefficient
8
) between the segmented regions from
projection and central reconstruction images. Such an analysis of clinical images is, however, limited by the lack of
Medical Imaging 2010: Physics of Medical Imaging, edited by Ehsan Samei, Norbert J. Pelc,
Proc. of SPIE Vol. 7622, 76220F · © 2010 SPIE · CCC code: 1605-7422/10/$18 · doi: 10.1117/12.845299
Proc. of SPIE Vol. 7622 76220F-1
Downloaded From: http://spiedigitallibrary.org/ on 07/15/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx