New Frontiers in Radiology Powered by Innovative Imaging

Ingrid Hein

November 19, 2018

CHICAGO — Radiology is on the brink of a new era, with an explosion of new image-assessment tools that use artificial intelligence, machine learning, and big data.

"Immune checkpoint inhibitors are fundamentally changing the way we practice radiology," said Julius Chapiro, MD, from the Yale School of Medicine in New Haven, Connecticut. And the new generation of drugs is changing the way the immune system reacts, he added.

"We are truly at a breakthrough point," Chapiro told Medscape Medical News. This will be reflected at upcoming the Radiological Society of North America (RSNA) 2018 Annual Meeting, where there will be "a lot of hype" about many of the clinical trials presented.

"For example, the world of cancer is experiencing a true revolution in immune-oncology," he explained. "In the past 2 years, we've seen hundreds of new clinical trials with novel activations of immune systems revealing possible ways to cure cancer." In these, personalized medicine plays an integral role, he pointed out.

Molecular imaging has the potential to guide more personalized therapy, which means radiologists need to find novel ways to assess tumor response.

We are truly at a breakthrough point.

With current standards, "we cannot necessarily reliably quantify changes," Chapiro said. Enlarged size is sometimes indicative of "pseudo progression. But it's important to look more closely at imaging features and characteristics to understand what is changing."

Results from the CheckMate trials looking at lung cancer are highly anticipated, he said.

"Immunotherapy takes the brakes off the immune system and allows it to fight against cancer cells that normally get turned off," said David Kim, MD, from UWHealth in Madison, Wisconsin, who is chair of the RSNA scientific committee.

As more immunotherapies are approved for the treatment of advanced cancer, radiologists are looking for ways to modify existing models so that clinical cancer imagers can work alongside oncologists, Kim told Medscape Medical News.

Among the studies that will be presented are an examination of the use of Ferumoxytol-enhanced MRI to explore tumor-associated macrophages and unmet imaging needs in immunotherapy drug development.

Discussion of the role AI plays in radiology is proliferating.

Artificial Intelligence, or Deep Learning, Dominates Abstracts

"The abstracts touching on some aspect of machine learning have skyrocketed this year," Kim reported. The informatics scientific committee got three times the number of abstracts in this field as it did last year.

Although the AI space is still mostly focused on whether a machine can recognize a problem — for example, identifying a nodule — radiologists are trying to determine whether it can make a diagnosis, he added.

The annual oration in diagnostic radiology — entitled Artificial Intelligence, Analytics, Informatics: The Future Is Here — will be delivered by Michael Recht, MD, from the NYU Langone Medical Center in New York City.

"Radiologists are seeking answers about what AI can offer," Kim explained. "Are there more efficiencies to be had? Can we increase our work flow? Can we improve?"

In addition to AI, radiomics — the extraction of data from medical images using algorithms to uncover disease characteristics and make predictions — will be in the spotlight.


"People are trying to find ways to predict response to therapy through imaging," Kim explained, such as making a cancer prognosis on the basis of imaging.

"We start with things we can see with the human eye and report various findings and morphologic changes. Sometimes this gives you a sense if a patient will do better or worse in the future," he pointed out. But you cannot quantify information you cannot see.

"To really look at textural differences and patterns, you need a computer to crunch the data," he pointed out.

Other New Technologies

New PET technologies, including non-FDG PET tracers, for molecular brain imaging will also be a hot topic. And there will be a presentation on PET agents for the molecular imaging of heart diseases, Kim reported.

There will also be a session on fast musculoskeletal imaging and other accelerated techniques that will delve into the way machine learning can be used to accelerate MRI.

This year, there will be 431 education courses. "The RSNA organizing committees have done a great job of marrying science and education," Kim told Medscape Medical News. Every subspecialty committee has a few of these planned.

"Many people with a community practice come to the meeting for education," he explained. Typically, education sessions include a didactic lecture, three or four scientific abstracts pertaining to that area, and then another didactic lecture. "The GI section has three of these sessions planned," he said.

There will also be a little friendly competition on the education side this year. In the Discovery Theater at the McCormick Center, teams of residents will compete to correctly diagnose images.

Follow Medscape on Twitter @Medscape and Ingrid Hein @ingridhein


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: