Despina Kontos PhD, Presenter: Nothing to Disclose

Jae Young Choi DPhil, Abstract Co-Author: Nothing to Disclose

Brad M. Keller PhD, Abstract Co-Author: Nothing to Disclose

Emily F. Conant MD, Abstract Co-Author: Scientific Advisory Board, Hologic, Inc

Andrew D.A. Maidment PhD, Abstract Co-Author: Research support, Hologic, Inc Research support, Barco nv Spouse, Employee, Real-Time Radiography, Inc Spouse, Stockholder, Real-Time Radiography, Inc

Accurate breast density estimation is becoming increasingly important for personalized breast cancer screening recommendations. We investigate the feasibility of obtaining reliable density measures from digital mammograms acquired at reduced radiation dose.

Bilateral, raw digital mammography (DM) images (Selenia, Hologic Inc.) from the ACRIN 4006 trial were retrospectively analyzed. For each exam, one set of DM images was acquired with standard dose, while another was acquired with about 15% reduction relative to the standard dose, per ACRIN 4006 protocol. Breast density was measured quantitatively with fully-automated FDA-cleared software (Quantra™ v.2.0, Hologic, Inc.), including BIRADS density, area percent density (PD%), and volumetric percent density (VD%). The agreement between BI-RADS density categories from standard-dose and low-dose images was estimated using Cohen’s weighted kappas. Linear regression and the Pearson correlation were used to test agreement for the continuous PD% and VD% estimates between standard-dose and low-dose mammograms.

Seventy-four image sets, each from a different study participant, were available for analysis (a total of 592 bilateral CC and MLO images). High agreement (k=0.87, 95% CI 0.83-0.91) was observed for BIRADS density estimates from standard-dose and low-dose mammograms. The correlation between standard-dose and low-dose PD% and VD% was r =0.94 (p <0.001, 95% CI 0.93-0.96) and r =0.95 (p <0.001, 95% CI 0.94-0.96), respectively. The slope, intercept, and goodness-of-fit R2 values of the linear regression for PD% were 0.95±0.01, 1.41±0.48, and 0.90±0.04, respectively; while for VD% were equal to 0.92±0.02, 0.86±0.23, and 0.90±0.02, respectively.

Our results show strong agreement between density estimates from standard-dose and low-dose digital mammograms acquired at a 15% reduction of radiation dose. This suggests that reliable density estimation is feasible from images obtained at reduced dose. Larger studies are needed to validate these findings at lower radiation dose levels.

Reliable density estimates at lower radiation dose could be used for purposes such as evaluating response to risk-reduction interventions and triaging women to the most appropriate screening modality

Kontos, D, Choi, J, Keller, B, Conant, E, Maidment, A, Effect of Reduced Radiation Dose on Breast Density Estimation in Digital Mammography: Data from the ACRIN 4006 Trial.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13044363.html