Sabee Y. Molloi PhD, Presenter: Research Consultant, Koninklijke Philips Electronics NV

Huanjun Ding, Abstract Co-Author: Nothing to Disclose

Elin Moa MS, Abstract Co-Author: Employee, Koninklijke Philips Electronics NV

Stephen Albert Feig MD, Abstract Co-Author: Medical Advisory Board, Hologic, Inc

Breast density is a significant risk factor in developing breast cancer. Breast density is currently reported by radiologists using BI-RADS categories, which is known to have a large inter-observer variability. The purpose of this study is to evaluate spectral mammography for quantification of breast density.

Four-view mammograms for 93 women from a previous study using a prototype Philips MicroDose Mammography SI system were used and all images were arranged in a random order for a blind comparison study. The system uses a photon counting detector to acquire energy resolved images in a single exposure with no additional dose to the patient. Four-category BI-RADS rankings were assigned independently by 10 radiologists based in the USA and the UK. The four-category ranking was converted to breast density scales (0 ~ 100%) using a linear relationship. Area-based breast density measurements were also performed by a physicist using Cumulus 4 and by an automatic image segmentation method based on a fuzzy C-mean clustering (FCM) technique. Volumetric breast density was calculated with a dual energy decomposition technique using the available spectral information. For all four techniques, the linear regression analysis was performed to investigate the correlation of the breast densities from the right and left breasts.

The breast densities from the right and left breasts showed a reasonably good linear correlation. The normalized variance about the best-fit line, which reflects the precision of the techniques, was estimated to be 8.4, 13.4, 6.1, and 1 for BI-RADS, Cumulus, FCM, and dual energy decomposition, respectively. This indicates that the variability in estimation of breast density is substantially lower using spectral mammography as compared with using BI-RADS, Cumulus and FCM in conjunction with standard mammography.

Spectral mammography may offer quantification of volumetric breast density with excellent precision during standard screening mammography. This will largely eliminate the inter- and intra-observer variability in the currently used BI-RADS ranking.

Breast density quantification is useful for assessing the risk of developing breast cancer. Spectral mammography can be used for accurate quantification of breast density in screening mammography.

Molloi, S, Ding, H, Moa, E, Feig, S, Volumetric Breast Density Quantification Using Spectral Mammography.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13026018.html