In an exciting overcrowded seminar, delegates gathered to listen to experts discuss how artificial intelligence (AI) and machine learning (ML) are changing radiology during one of Tuesday’s plenary sessions at RSNA 2017 in Chicago last week.
As one of the most tending subjects in radiology today, it’s no wonder why a high number of radiologists, physicists and vendors of AI tools all populated the room. However, presenters made it abundantly clear that radiologists need to use the technologies as daily tools for diagnosis and treatment efficiently in order for AI to function successfully.
During the session "Harnessing Artificial Intelligence," Dr. Keith Dreyer, vice chair of radiology and director of the Center for Clinical Data Science at Massachusetts General Hospital, Boston, and chair of the American College of Radiology's Commission on Informatics, explained to the audience the complexities involved in teaching computers to read images.
"Machines are getting smarter faster than people are," said Dr. Dreyer.
Dr. Dreyer explained that radiology needs globally accepted ways to develop and incorporate AI, in order to make it easy for developers to create new applications and integrate them into imaging devices and clinical information systems. Dr. Dreyer referred to the issue of developing AI as “a "healthcare AI ecosystem."
During the session, Dr. Antonio Criminisi, a principal researcher at Microsoft in Cambridge, U.K., also referred to his own research and Microsoft’s InnerEye project (using medical imaging AI to empower clinicians), to explain how these smart technologies should only be thought of as powerful devices to enhance productivity, rather than to solely rely on.
"This technology is not for doing things that you already know how to do well, but to do things that you wish you were able to do," Dr. Criminisi said.
Perfecting AI is not the priority here, as Dr. Dreyer explained, rather it is about getting it to work efficient enough for daily use.
Concluding his remarks, Dr. Dreyer raised an interesting question for the audience. "Do you need AI to be at the top? Or would it be enough for it to be somewhere in the middle, which is easier to achieve? I would argue there are many places on the globe where it would be adequate to have a good solution."