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
Conclusion
Our results demonstrated a high level of agreement between sequential mammograms taken on Hologic and GE,
or Fuji CR and Fuji DR systems using Volpara™ and support the use of the software for determining temporal changes in breast density
Density maps generated from the “For Processing” images standardizes the mammograms from different vendors,
allowing for a better judgement of temporal changes.
A comparison of our temporal density map movie with the movie using the “For Presentation” mammograms emphasizes how important it is to normalize the images first,
to take into account the different processing algorithms of each vendor.
It also highlights the benefits of applying image registration and feature mapping methods before appending sequential mammograms into a temporal movie.
The pixel mapping normalizes the breast sizes and positioning on the mammogram,
and also allows them to be overlayed for the movies.
Breast density is one of the few modifiable breast cancer risk factors and several studies have now drawn attention to the potential clinical applications of accurately tracking changes in breast density over time [4,
6,
9].
However,
the “For Presentation” images that radiologists use to assess breast density are not reliable for monitoring changes over time.
We have outlined an objective method for transforming sequential mammogram images into temporal movies,
which can be used on any digital mammographic images regardless of the x-ray system used,
or the imaging conditions.