Radiologists have traditionally determined breast density by visually comparing the amount of fibroglandular tissue (white/light) vs. fatty (gray) areas on a mammogram. The radiologist typically reports breast density in one of four Breast Imaging Reporting and Data System (BI-RADS®) categories: A) fatty, B) scattered fibroglandular density, C) heterogeneously dense, or D) extremely dense. Breasts that are heterogeneously dense or extremely dense are considered “dense breasts.” Visual determination of density is a subjective assessment, which can vary between radiologists, especially for breasts that are on the borderline between scattered fibroglandular and heterogeneously dense.
Computer software can be used to quantify and characterize breast density on a mammogram. Any density assessment (radiologist/visual or radiologist + automated tool) should be reassessed on each examination as some breasts may decrease in density and become fatty replaced (and no longer dense) as a woman ages. Conversely, breast density may increase if a woman starts using menopausal hormone supplements. Breast density may also fluctuate with weight gain or loss, appearing denser with weight loss and less dense with weight gain (note: the amount of fat changes but the amount of dense tissue remains the same). Density assessment is used when considering supplemental screening. BI-RADS density category is also used in some models that estimate future risk of developing invasive breast cancer.
What is it?
Software that can calculate the relative amount of dense tissue in the breast. Some software also considers if particular areas of the breast are dense (i.e., masking potential). The density assessment data are obtained by evaluating a digital mammogram or digital breast tomosynthesis exam with sophisticated algorithms and can be made available for the radiologist when reading the mammogram. Some software products can only be used at the time of acquiring the mammogram when all the “raw data” are available.
How it works
Automated assessments calculate density as either area (length x width) or volume (length x width x thickness) percent density. The assessed area or volume of dense tissue is divided by the area or volume of the entire breast and then multiplied by 100 to yield a percentage. This percentage is generally then correlated to one of the four BI-RADS categories: fatty, scattered fibroglandular density, heterogeneously dense, or extremely dense. We do not fully understand whether the absolute amount of dense tissue or the percent of dense tissue is more important, though an analysis of multiple methods [1, 2] found percent density more predictive of risk of developing breast cancer. Complexity of the texture [3, 4] of the breast tissue may be more important in predicting risk than the amount of dense tissue, though further study is needed. Artificial intelligence trained on mammograms can augment conventional risk assessment that includes breast density [5].
Benefits
Provides consistent breast density assessment and quantification across all patient populations. Removes the inter- and intra-radiologist subjectivity and variability of visual assessment of breast density.
Considerations
Software is costly and is not separately reimbursed by insurance carriers nor are radiologists required to use it. Automated breast density assessment may be affected by positioning of the breast during the mammogram and can produce erroneous readings in almost entirely fatty or extremely dense breasts. There are differences in software technology. Some approaches calculate the amount of fibroglandular tissue, some calculate percent area or percent volume that is dense, and some also consider the texture and variability (complexity) of density within the breast. Automated software typically provides one “average” measurement across the whole breast, or the maximum density value of the left or right breast. If one quadrant of the breast is particularly dense, this may prompt supplemental screening by the radiologist even though the software average density score across the entire breast may not be dense: any computerized measurement of density should be reviewed in the context of a particular woman’s mammograms. Visual density assessment and software approaches show similar performance in identifying risk of screen-detected and interval cancer [6, 7]. Further investigations are needed on the utility of volumetric methods in screening settings, including their contribution to specific approaches tailored to a woman’s risk.
1. Pettersson A, Graff RE, Ursin G, et al. Mammographic density phenotypes and risk of breast cancer: a meta-analysis. J Natl Cancer Inst 2014; 106
2. Eng A, Gallant Z, Shepherd J, et al. Digital mammographic density and breast cancer risk: A case inverted question markcontrol study of six alternative density assessment methods. Breast Cancer Res 2014; 16:439
3. Kontos D, Winham SJ, Oustimov A, et al. Radiomic phenotypes of mammographic parenchymal complexity: Toward augmenting breast density in breast cancer risk assessment. Radiology 2019; 290:41-49
4. Mainprize JG, Alonzo-Proulx O, Alshafeiy TI, Patrie JT, Harvey JA, Yaffe MJ. Prediction of cancer masking in screening mammography using density and textural features.Acad Radiol 2019; 26:608-619
5. Yala A, Lehman C, Schuster T, Portnoi T, Barzilay R. A deep learning mammography-based model for improved breast cancer risk prediction. Radiology 2019:182716
6. Destounis S, Johnston L, Highnam R, Arieno A, Morgan R, Chan A. Using volumetric breast density to quantify the potential masking risk of mammographic density.AJR Am J Roentgenol 2017; 208:222-227
7. Kerlikowske K, Scott CG, Mahmoudzadeh AP, et al. Automated and clinical breast imaging reporting and data system density measures predict risk for screen-detected and interval cancers: A case-control study. Ann Intern Med 2018; 168:757-765