By Prof. Nico Karssemeijer
The reading by radiologists of the mammograms generated in current breast screening programs is an extremely time-consuming process, a situation which
is likely to be exacerbated by the increasing use of digital breast tomosynthesis systems in the future. In addition, even with the currently recommended
process of using double readers, there are still a significant number of breast cancers which are not identified in screening mammography. Several years ago, it
was thought that the use of Computer-Aided Detection (CAD) systems could address these issues, but in practice such CAD systems failed to live up to expectations.