Over 90 Total Lots Up For Auction at One Location - WA 04/08

Mayo Clinic and Nvidia partner on Deep Learning Institute at RSNA

Press releases may be edited for formatting or style | November 26, 2018 Artificial Intelligence
ROCHESTER, Minn. — Artificial intelligence (AI) is going to change health care, including the practice of radiology, profoundly. But rather than machines taking over, they will work collaboratively with clinicians and researchers to improve the care patients receive.

“If somebody puts their head in the sand and wakes up and pulls their head out five years later, the practice will be very different,” says Bradley Erickson, M.D., Ph.D., professor of radiology.

To help with this transitional time, Dr. Erickson and colleagues in the Radiological Society of North America and Nvidia Corp., a computer chip manufacturer and technology company, developed a course for radiologists interested in acquiring or developing the skills needed to navigate this new technology. This course, called the “Deep Learning Institute,” will be held throughout the Scientific Assembly and Annual Meeting Nov. 25–30 in Chicago.

There are an introductory course and four others covering different aspects of deep learning — four 90-minute courses daily Nov. 25–29, and two Nov. 30. Participants will be expected to bring a laptop and write code.

“We will give them most of the code, but we have intentionally introduced some bugs to see if they can find them and if they really understand what’s going on,” Dr. Erickson says. “That is definitely a hands-on course, and the intent is to give them enough experience with deep learning that they can go home and have essentially a good starter set where they can start to do some of this work at home.”

The advances in AI and machine learning are not unlike the emergence decades ago of VisiCalc, a predecessor to Excel spreadsheets, Dr. Erickson says. He adds that people at the time assumed accountants would no longer have a use — and the profession itself would no longer exist.

“There are more accountants today than back in 1978 when VisiCalc came out, and they get paid better, even accounting for inflation, than when VisiCalc came out,” he says. And as with VisiCalc, artificial intelligence will take over the “mundane, adding-machine-type work that we do in radiology” and leave the roles that cannot be automated to the humans.

“I think ultimately we still want a human with their hands on the medical “steering wheel,” just like with self-driving cars,” Dr. Erickson says. “I think that this is why it would be unwise for a radiologist to say, ‘AI doesn’t matter.’ I think it will change what we do but I think that if we remain engaged it will actually make the practice of radiology a lot better and a lot more interesting.”

You Must Be Logged In To Post A Comment