Gautam S Muralidhar MS, Presenter: Nothing to Disclose

Mia Kathleen Markey PhD, Abstract Co-Author: Researcher, Hologic, Inc

Alan C Bovik, Abstract Co-Author: Researcher, Hologic, Inc

Tanya Stephens MD, Abstract Co-Author: Nothing to Disclose

Tamara Haygood MD, PhD, Abstract Co-Author: Nothing to Disclose

William Robert Geiser MS, Abstract Co-Author: Nothing to Disclose

Naveen Garg MD, Abstract Co-Author: Nothing to Disclose

Gary J. Whitman MD, Abstract Co-Author: Nothing to Disclose

We evaluated the feasibility of characterizing breast masses as seen on breast tomosynthesis projection images using a stereoscopic (stereo) display.

Tomosynthesis data were provided by Hologic, Inc. (Bedford, MA, USA). Each case was comprised of 15 projections that spanned an angular range of 15 degrees. Two projection images that were ± 4 degrees apart from the zero angle projection were chosen to form a stereo pair. In this study, we used only the craniocaudal stereo pairs from 47 cases. The images had their contrast enhanced and were displayed on a Planar PL2010M stereo display (Planar Systems, Inc., Beaverton, OR, USA). The 47 cases were comprised of 23 biopsy proven masses and 24 normals. An experienced breast imager who did not participate in the observer study verified the truth for each case stereoscopically. Two experienced breast imagers participated in a blinded observer study as readers after passing the Randot stereo acuity test. The readers were asked to describe the most suspicious abnormality using the BI-RADS® (American College of Radiology) descriptors. The readers were also asked to describe the lesion subtlety on a scale of 1-5. The AC1 and AC2 (weighted AC1) coefficients were used as alternatives to the Kappa coefficient to assess the inter-reader agreement.

The sensitivities achieved by the two readers in detecting the correct abnormality were 86.9% and 91.3%, while the specificities achieved were 79.1% and 83.3%. Good inter-reader agreement was observed for BI-RADS® mass margin and assessment ratings (AC1 = 0.64 and 0.73, respectively), while excellent inter-reader agreement was observed for mass subtlety ratings (AC2 = 0.94). Moderate agreement was observed for BI-RADS® mass shape ratings (AC1 = 0.28).

Breast masses can be reliably characterized on breast tomosynthesis projections with stereoscopic viewing.

Breast tomosynthesis projections are amenable to stereoscopic viewing, which has the potential to aid in image interpretation and treatment planning.

Muralidhar, G, Markey, M, Bovik, A, Stephens, T, Haygood, T, Geiser, W, Garg, N, Whitman, G, Stereoscopic Characterization of Breast Masses on Tomosynthesis Projection Images.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11008539.html