Abstract Archives of the RSNA, 2012
Michiel Kallenberg, Presenter: Nothing to Disclose
Carla H. van Gils PHD, Abstract Co-Author: Research Grant, Bayer AG
Ritse Maarten Mann MD, PhD, Abstract Co-Author: Nothing to Disclose
Nico Karssemeijer PhD, Abstract Co-Author: Shareholder, Matakina International Limited
Scientific Board, Matakina International Limited
Shareholder, QView Medical, Inc
Research Grant, Riverain Medical
The association between breast density and breast cancer risk has been established with semi-automated, area based measurement of breast density. In this work we investigate the performance of a fully automated, volumetric method to predict breast cancer risk and the risk of a false positive referral.
The data that was used in this work comprised 50,446 digital screening exams from 33,029 women in the Dutch Breast Cancer Screening Program. The images were recorded on a Hologic Selenia FFDM system, using standard clinical settings. In total 1145 exams were referred for further assessment. 844 exams of these 1145 referrals turned out to be false positives. 301 referrals were biopsy proven malignant breast cancers.
For each mammographic image we assessed volumetric breast density with commercially available software (Volpara®, Matakina Technology, Wellington, New Zealand). We examined the relationship between volumetric breast density estimates and diagnostic outcome by calculating odds ratios and their 95% confidence intervals for percent density converted to the Volpara density grade (VDG), which is a four point scale analog to BI-RADS density scores.
We used published data on sensitivity of mammography screening to analyze the potential effect of interval cancers on risk association, as the actual frequency of interval cancers in our dataset in not yet known.
After adjusting for age, the odds-ratio for a screen detected breast cancer is 1.53 (0.91-2.68 (95% CI)) comparing highest breast density to low breast density (i.e. VDG 4 vs 1). If we include the estimated contribution of interval tumors, the risk for breast cancer in dense breasts is between 1.49 (1.20-3.04) and 3.58 (2.46-5.31). Breast density decreases the positive predictive value of a referral. The age adjusted odds-ratio for a false positive referral is 2.83 (2.03-4.00) comparing high breast density to low breast density.
High volumetric breast density, as assessed by a robust, automated method, is associated with both breast cancer and false positive referrals.
Volumetric breast density estimation may be used for automated selection of women at high risk for breast cancer and false positive referrals, and may hence offer opportunities for tailored screening.
van Gils, C,
Association between Automated, Volumetric Measures of Breast Density and Diagnostic Outcome of Mammography Screening Examinations. Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12027520.html