Artificial intelligence for imaging breast cancer

Abstract

The incidence of breast cancer in women is high and has been increasing over the years. However, the mortality rate is decreasing as a result of improved early detection and treatment. These findings have driven worldwide efforts to increase breast cancer awareness and to deliver early cancer detection with improved treatment possibilities. The achievements obtained for early detection would not have been possible without the massive strides in development of information technology and computational power. Where only 2D analogue images were available 20 years ago, today we consistently work with 3D imaging. These technologies however have resulted in an exponential increase in data volume that consistently challenges radiologists’ capabilities. Artificial Intelligence promises to be a potential solution for the current imbalance between the demand on radiologists’ time and the increasing volume of imaging data generated that needs to be reviewed. Artificial Intelligence has further potentials for improving future practice workflows. Ongoing studies are working towards creating patient-specific screening and diagnostic workups, improving patient scheduling and protocol selection, radiation dose reduction, automated acquisition, improved image reconstruction to increase lesion visibility, a prioritised reading list (putting highly suspicious cases first and non-suspicious cases last), decreasing reading time, and prediction of breast cancer risk. Recent studies have shown that artificial Intelligence can improve radiologists’ cancer detection rates, reduce recall rates, and reduce reading times of high data volume exams such as 3D mammograms.