This content is part of the Conference Coverage: RSNA 2018 conference coverage and analysis

Conference Coverage

Browse Sections

RSNA 2018 shines light on AI, machine learning

RSNA 2018 doesn't stray far from last year's themes, so conference-goers should expect to be inundated with all the ways that artificial intelligence and machine learning will modernize the medical imaging field.

AI and machine learning in healthcare took center stage last year at one of the premier medical imaging shows....

And it looks like this year will be no different.

The Radiological Society of North America (RSNA) will hold its annual meeting next week. And like everything else in tech, AI is expected to be a major focus. Billed as the largest healthcare event to take place in North America, RSNA 2018 promises to attract thousands of radiologists, physicians and other medical professionals.

Together, conference-goers will hear how AI and machine learning are upending the medical imaging space from the likes of Arie Meir, product manager for Google Cloud; Fei-Fei Li, director of the Stanford Artificial Intelligence Lab and Stanford Vision and Learning Lab; and George Shih, M.D., chairman of RSNA's informatics subcommittee and associate professor and vice chairman for informatics in Weill Cornell Medical College's radiology department.

"AI will be really hot again this year," Shih put simply.

Machine learning at RSNA 2018

RSNA 2018 will feature hundreds of exhibits, many of which will highlight machine learning. That's no surprise to Shih, who said he's seen the volume of machine-learning-related abstracts increase threefold from last year.

George Shih, M.D., chairman of RSNA's informatics subcommittee and associate professor and vice chairman for informatics in Weill Cornell Medical College's radiology departmentGeorge Shih

"The quality of these abstracts has gotten better, and we've increased the number of AI scientific sessions to try to accommodate all these great abstracts," he said. 

Shih recommended attendees take a look at the machine learning showcase in the technical exhibit area, which he said has been significantly expanded to accommodate additional vendors. There, attendees can get caught up on the latest product developments and software in machine learning and AI for medical imaging.

Deep learning, a form of machine learning, will also get some attention. Sessions will be held on how deep learning is having an impact on medical imaging, including breast, prostate and gastrointestinal imaging.

AI at RSNA 2018

AI will be really hot again this year.
George ShihM.D., chairman of the RSNA radiology informatics subcommittee

Shih said attendees should expect AI to be a prominent theme this year, with nearly every vendor showcasing an AI-powered product in the exhibition hall.

He said attendees should keep an eye on nonpixel research and commercial algorithms, which he believes will be an emerging trend at RSNA 2018. Historically, the excitement around AI has been based on its ability to find patterns in imaging or pixel data, such as an algorithm detecting a disease from an X-ray. Now, new techniques allow AI to be used for other data, such as reports, Shih said.

"Not only are we, as a field, building algorithms for X-rays, but we're also doing it for nonpixel data in radiology," Shih said.

Shih is also looking forward to the RSNA Pneumonia Detection Challenge award ceremony, which will take place Monday at the machine learning showcase. Participants were tasked with building a machine learning algorithm that could automatically detect pneumonia in medical images. To develop a pneumonia data set for the challenge, RSNA collaborated with the U.S. National Institutes of Health; the Society of Thoracic Radiology (STR); Kaggle, a data science competition platform; and, a medical imaging platform.

The top 10 winners will present their algorithms during the conference, Shih said.   

"All top 10 algorithms and the RSNA/STR-annotated pneumonia data set will be open-sourced and available for anyone to use, which I think will be a significant contribution to medical imaging AI," Shih said.  

Dig Deeper on Healthcare IT systems and applications

Cloud Computing
Mobile Computing