Machine Learning in Cardiac CT
DOI: 10.1007/s40134-017-0241-9 Scott P. Landreth James V. Spearman
Purpose of Review
This review covers the basic principlesof machine learning (ML) and current applications in thesubspecialty o f cardiac imaging at computed tomographyin diagnostic radiology.
This review covers recent publicationsfor automated image processing, diagnostic and prognosticsupport, as well as novel integrations of ML into extantimaging applications in advanced cardiac computedtomography. Where available, ML algorithms are com-pared to current gold standards and descriptions of thenature and value of the advances are described.
Machine learning in clinical imaging is consid-ered by many to represent one of the most promising areasof research and development in diagnostic radiology.Sophisticated Machine Learning systems like IBM’s Wat-son have capt ured the public’s attention in their ability tomimic human capacity for pattern recognition in extremelylarge data sets. There are numerous recent research publi-cations utilizing Machine Learning algorithms to eitherautomate processes or improve diagnosis or even createentirely new forms of evaluation previously considered outof reach for cardiac CT imaging