Abstract Archives of the RSNA, 2014
SSA20-02
Fusing Microwave Radar Imaging with Digital Breast Tomosynthesis for High Contrast, High Resolution Breast Cancer Detection
Scientific Papers
Presented on November 30, 2014
Presented as part of SSA20: Physics (Non-Conventional Techniques)
Carey Milford Rappaport PhD, Presenter: Nothing to Disclose
Matt Tivnan, Abstract Co-Author: Nothing to Disclose
Emma Kaeli, Abstract Co-Author: Nothing to Disclose
Richard Obermeier, Abstract Co-Author: Nothing to Disclose
Richard H. Moore, Abstract Co-Author: Research support, General Electric Company
Research support, Siemens AG
Research support, Astrophysics Inc
Research support, SAP AG
Daniel B. Kopans MD, Abstract Co-Author: Research Grant, General Electric Company
Royalties, Cook Group Incorporated
Consultant, Siemens AG
Jose Angel Martinez Lorenzo PhD, Abstract Co-Author: Nothing to Disclose
Simulated microwave propagation employing DBT-derived gross tissue regions (priors) as a background provides a means of reducing clutter from the random fibro-glandular tissue distribution, which in turn allows spatial localization of a small tumor. The ~10% dielectric contrast difference between the simulated tumor and the healthy fibroglandular tissue generates a relatively strong microwave signal once the clutter is removed.
Digital Breast Tomosynthesis (DBT) mammography detects ~70% of cancers that will become evident. Tissue dielectric contrast may permit detection of otherwise undetected tumors early enough to intervene.The dielectric contrast between cancerous tissue and healthy high water content breast tissue is an order of magnitude greater than X-ray contrast, but microwave imaging resolution is significantly poorer than x-ray imaging.
Using a DBT image from a health breast, forward FDFD microwave simulations were computed for various transmission locations. A second set of computational simulations were computed for the same DBT image, but with an added modeled clinically-relevant tumor volume. The computed simulations indicated a difference in the vacinity of the true tumor location with scattering intensity that was 10-3 times the response of the measured signal. This tumor response was a systematic, organized pattern, with no artifacts or clutter.
Employing a downsampled DBT three dimensional reconstruction of the phsical tissue dielectric distribution as the background geometry improves the accuracy of computational EM finite difference frequency domain (FDFD) scattering simulation. The scattered field distributions of healthy breasts and diseased breasts with simulated lesions can be predicted with sufficient fidelity to distinguish differences. By digitally subtracting the modeled, assumed healthy background field from the measured field, inconsistencies generated by the simulated cancerous lesion are highlighted, and the tumor can be detected and accurately located in three dimensions.
Rappaport, C,
Tivnan, M,
Kaeli, E,
Obermeier, R,
Moore, R,
Kopans, D,
Martinez Lorenzo, J,
Fusing Microwave Radar Imaging with Digital Breast Tomosynthesis for High Contrast, High Resolution Breast Cancer Detection. Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL.
http://archive.rsna.org/2014/14018670.html