Said Pertuz PhD, Presenter: Nothing to Disclose

Elizabeth McDonald MD, PhD, Abstract Co-Author: Nothing to Disclose

Susan Weinstein MD, Abstract Co-Author: Nothing to Disclose

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

Despina Kontos PhD, Abstract Co-Author: Nothing to Disclose

Accurate breast density estimation is important for breast cancer risk assessment and guiding personalized breast screening recommendations. We investigate the feasibility of fully-automated volumetric breast density estimation (VBD) from digital breast tomosynthesis (DBT), and compare to VBD estimates from digital mammography (DM) and breast MRI. Compared to 2D mammography, DBT visualizes the 3D distribution of fibroglandular tissue, having the potential to allow for more accurate VBD estimation.

Bilateral DBT images, DM images (Selenia, Hologic Inc.) and sagittal MRI scans (GE LX echo speed, Siemens) were retrospectively collected from 63 women undergoing breast cancer screening within the course of one year (2010-11). A fully-automated algorithm was developed to segment the fibroglandular tissue and measure VBD from all DBT images. The proposed algorithm exploits the geometry of the acquisition of DBT sequences as well as the relationship between image intensity and tissue density and achieves 3D segmentation of the fibroglandular tissue by analyzing both the projection images and reconstructed DBT slices. For comparison, the DM images were processed with FDA-cleared software (Volpara 1.5, Matakina) and the MR images were processed with previously validated automated software to obtain corresponding VBD estimates. The Pearson’s correlation and linear regression were used to compare the obtained multi-modality VBD estimates.

Substantial agreement is observed between bilateral VBD estimates from DBT images (r = 0.89, 95% CI: 0.83-0.93, p<0.001). Estimates of the total breast volume and percent volumetric breast density from DBT are highly correlated with DM with r = 0.99 (95% CI: 0.98-0.99) and r = 0.88 (95% CI: 0.81-0.93); as well as with the MR-based estimates with r = 0.95 (95% CI: 91-0.96) and r = 0.76 (95% CI: 63-0.85), respectively (p<0.001). Corresponding correlations between DM and MRI are r = 0.95 (95% CI: 0.92-0.97) and r = 0.73 (95% CI: 0.59-0.83).

Fully-automated 3D fibroglandular tissue segmentation and VBD estimation from DBT images is feasible and shows strong agreement with existing volumetric techniques based on DM and MRI images.

Fully-automated quantitative VBD estimation from DBT could result into more accurate measures of the fibroglandular tissue in the breast and ultimately more accurate measure of breast cancer risk.

Pertuz, S, McDonald, E, Weinstein, S, Conant, E, Kontos, D, Fully Automated Volumetric Breast Density Estimation from Digital Breast Tomosynthesis Images: Multi-modality Comparison with Digital Mammography and Breast MRI.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14015042.html