Local doctor testing cancer-screening AI software
Using an algorithm, software can detect malignant and nonmalignant breast cancer tumors 7%-8% higher than humans can.
A local doctor is testing a new artificial intelligence product that he said significantly improves breast cancer detection.
Dr. Mark Traill, a radiologist at Metro Health - University of Michigan Health, is one of the first in the state to use FDA-approved ProFound AI, which uses technology he said will be “massively disruptive” in the field.
The software helps doctors spot breast cancer tumors more accurately, reducing false positives and unnecessary patient callbacks for additional testing.
Using an algorithm, the software can analyze each 3D mammography image for a patient and detect malignant and nonmalignant tumors at a rate of about 7%-8% higher than humans can, Traill said.
“The algorithm can look at a scan and put together patterns that you can't even see,” he said.
Traill said it’s easy for even several doctors to completely overlook abnormalities in a screening.
“Nobody finds them all,” he said.
When New Hampshire-based medical technology company iCAD rolled out the technology last year, Forbes called it one of the 2018’s biggest moments in AI.
Through Traill’s relationship with iCAD, the company donated the technology to Metro Health a little over a month ago, which Traill alone is using on a research basis with the University of Michigan.
He said Metro plans to expand the use of the technology throughout the entire system this month. At that time, the hospital will be one of three systems in the state using it.
“I would expect that probably within a year or so, that number is going to be much, much higher,” Traill said.
Traill said he remembers the excitement when MRI technology was introduced because of the changes it brought in diagnoses.
“This is going to be a lot bigger than that,” he said.
“There's been a lot of hype, but I don't think it's overdone for this. I think this is something that is really going to have a huge impact on radiology and across the board in medicine.”
He said there was some apprehension from radiologists when the technology was first being introduced because of the fear that it will automate what the doctors do.
“That's not going to be the case,” he said. He argues that radiologists who use the technology no longer have to spend time reading mammograms but instead will be able to offer more valuable care.
“AI is not going to replace radiologists, but radiologists who use it and embrace it are going to replace those that don't.”
Traill said he has been attending conferences and following advances in this technology for about four years, and the progress has been “surprisingly fast” and seems to be somewhat outpacing its utilization for now.
The technology still is in the early stages. In the next decade, he said he expects it may even be able to diagnose issues, not just find abnormalities.