Computer Aided Detection (CAD) for Digital Mammography: A Retrospective Reading Study for Consideration on Utilizing CAD Most Effectively

A reading test was performed for a digital mammography (MMG)-CAD system and analyzed the type of readers that used CAD most effectively. The database for the reading test consisted of 40 breast cancer cases and 60 cases without breast cancer. 12 readers interpreted mammograms both with and without CAD. We divided the readers into 2 groups. Group A included readers who had either an extremely high or a relatively low ability of interpretation and insufficient understanding of CAD. Group B included the rest of the readers. As the results, in Group A, there was no statistical difference between the results with and without CAD. In Group B, there was a statistical difference between results with and without CAD in sensitivity, but specificity. This result implies that both sufficient reading training of MMG and the sufficient explanation and training of MMG-CAD is important.