U of C's Giger: How machine learning can succeed in medical imaging

In an editorial in the March issue of the Journal of the American College of Radiology, Maryellen L. Giger, PhD, and professor of radiology at the University of Chicago, discussed what must occur for machine learning to succeed in health imaging and what clinicians and patients should expect in the future from the synergy of medical imaging and artificial intelligence.

Risk assessment, detection, diagnosis and therapy response are a few examples of radiological imaging tasks that have advanced and benefited from the implementation of machine learning technology. 

"For deep learning in radiology [and health imaging] to succeed, note that well-annotated large data sets are needed since deep networks are complex, computer software and hardware are evolving constantly and subtle differences in disease states are more difficult to perceive than differences in everyday objects," Giger wrote. 

Specifically, Giger asserted that radiomics, "an expansion of computer-aided diagnosis,” can advance the field of health imaging by providing predictive image-based phenotypes of disease for precision medicine and quantitative image-based phenotypes for data mining for other areas of discovery, such as imaging genomics. 

Because she believes that imaging examinations will be routinely obtained in clinical practice because of machine learning, the opportunity to improve "decision support" in image interpretation will generate more efficiency for clinicians and overall better patient outcomes. 

"The term of note is decision support, indicating that computers will augment human decision making, making it more effective and efficient," Giger said. "The clinical impact of having computers in the routine clinical practice may allow radiologists to further integrate their knowledge with their clinical colleagues in other medical specialties and allow for precision medicine."