This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Cancer, Observer performance, Mammography, Breast
Authors:
K. Schilling, J. The, S. Griff, L. Oliver, R. Mahal, M. Saady, M. V. Velasquez; Boca Raton/US
DOI:
10.1594/ecr2015/C-1281
Aims and objectives
The American College of Radiology’s Breast Imaging Reporting and Data System (BI-RADS) is the predominant method for assessing mammographic density in the clinic.
When this study was conducted,
we were using the 4th Edition of BI-RADS.
This method involves radiologists visually assigning the breasts into one of four categories [1]:
- The breast is almost entirely fat (<25% glandular)
- There are scattered areas of fibroglandular densities (approximately 25-50% glandular)
- The breast is heterogeneously dense,
which could obscure detection of small masses (approximately 51-75% glandular)
- The breast tissue is extremely dense.
This may lower the sensitivity of mammography (>75% glandular)
Breast density is a strong,
independent risk factor for developing breast cancer,
and increasing breast density also decreases the sensitivity of mammography [2].
Despite the significant implications of breast density on screening mammography and the requirement in many states to inform women there is considerable inter- and intra-observer variability in radiologists’ determination of breast density [3,
4].
Even amongst American Board of Radiology examiners,
a recent study demonstrated wide variation in the BI-RADS density assessment between them,
with kappa scores ranging from 0.347 to 0.665 [5].
This study also demonstrated that technical factors,
such as the x-ray system vendor,
can influence visual breast density assessment.
To improve the accuracy and reproducibility of breast density assessment,
several new methods have been developed,
each with their own associated benefits and drawbacks.
For example,
semi-automated methods such as Cumulus,
can predict the risk of developing breast cancer,
but are too labour-intensive for widespread clinical use [6].
Fully-automated area-based methods,
such as ImageJ and AutoDensity,
are still being validated clinically,
but are limited by the fact that they are measuring a 3-dimensional phenomenon,
from a 2-dimensional projection [6,
7].
Fully-automated volumetric-based methods that measure the physical amount of fibroglandular tissue in the breast are reproducible,
proven to be associated with breast cancer risk,
FDA cleared and slowly coming into clinical use [8].
We sought to study whether the use of fully-automated volumetric breast density (VBD) software improves the visual BI-RADS inter-reader agreement between radiologists at our clinic.