Testing realism of software breast phantoms:  texture analysis of synthetic mammograms

Predrag R. Bakic; Brad Keller; Yuanjie Zheng; Yan Wang; James C. Gee; Despina Kontos; Andrew D. A. Maidment

doi:10.1117/12.2008173

19 March 2013 Testing realism of software breast phantoms: texture analysis of synthetic mammograms

Predrag R. Bakic, Brad Keller, Yuanjie Zheng, Yan Wang, James C. Gee, Despina Kontos, Andrew D. A. Maidment

Proceedings Volume 8668, Medical Imaging 2013: Physics of Medical Imaging; 866824 (2013) https://doi.org/10.1117/12.2008173
Event: SPIE Medical Imaging, 2013, Lake Buena Vista (Orlando Area), Florida, United States

Abstract

Software breast phantoms have been developed for pre-clinical validation of breast imaging systems. Realism is of great importance for the acceptance and the range of applications of breast phantoms. In this paper we have assessed the phantom realism based upon the analysis of mammographic texture properties. Texture analysis is of interest since it reflects the spatial tissue distribution, which is known to correlate with breast cancer risk. We compared texture properties of synthetic mammograms generated using software breast phantoms with clinical data. A total of 133 phantom images were synthesized using software phantoms developed at the University of Pennsylvania. The phantoms were designed using two different anatomy simulation methods: an octree-based recursive partitioning method and a region growing method. The synthetic images were generated assuming a clinically used acquisition geometry and mono-energetic x-ray beam with no scatter. The clinical data included 60 anonymized mammograms selected retrospectively from screening cases at the University of Pennsylvania. The same postprocessing was applied to clinical and phantom images. The texture analysis was performed using fully automated software which extracts a battery of features from analyzed images. The histograms of texture properties extracted from phantom images were compared with those from clinical mammograms, separately for the two anatomy simulation methods. The histogram agreement was quantified using symmetrized Kulback-Leibler divergence. We observed good agreement for most of the analyzed 25 features. In more than a half of the features, the octree-based simulation method yielded better agreement with clinical data as compared with the region growing method.

Citation Download Citation

Predrag R. Bakic, Brad Keller, Yuanjie Zheng, Yan Wang, James C. Gee, Despina Kontos, and Andrew D. A. Maidment "Testing realism of software breast phantoms: texture analysis of synthetic mammograms", Proc. SPIE 8668, Medical Imaging 2013: Physics of Medical Imaging, 866824 (19 March 2013); https://doi.org/10.1117/12.2008173

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available