Keywords:
Computer Applications-Detection, diagnosis, Imaging sequences, Mammography, Breast, Screening, Cancer, Quality assurance
Authors:
X. Lin, N. Sauber, R. Highnam; Wellington/NZ
DOI:
10.1594/ecr2013/C-1770
Purpose
Breast density is known to be significant predictor of breast cancer risk,
with studies reporting up to a 6-fold increase in the risk of developing cancer in women with very dense breasts compared to women with very fatty breasts [1-3].
In addition to single timepoint assessments of breast density,
studies have now suggested that abnormal changes in breast density over time could also be associated with an increased risk of developing breast cancer [4,
5]. As an example,
Vachon et al (2010) found that breast cancer risk was reduced by 28% for women whose breast density decreased by at least one BI-RADS category over approximately 6 years,
compared to women who experienced no change.
In addition to predicting breast cancer development,
temporal changes in breast density may also be useful in judging the effects of age,
body mass index (BMI),
and menopause on the breast,
as well as assessing the effectiveness or otherwise of certain drugs.
For example,
hormonal replacement therapies (HRT) and endocrine therapies (e.g.
tamoxifen) have also been shown to increase and decrease breast density,
respectively [6,
7].
Temporal changes in breast density could potentially be used to stop the drug treatment if it’s having an adverse effect on density [6,
8,
9].
To date,
no efficient method has been identified that can objectively assess temporal changes in breast density.
Limitations associated with current methods include intra- and inter-reader variability,
which is made even more difficult by the inherent differences in the “For Presentation” mammographic images generated on different vendor x-ray machines.
In this paper,
a novel,
objective method for transforming sequential mammograms into temporal density movies is introduced,
which addresses some of these issues.