The tool helped clinicians correctly identify up to six more aneurysms in 100 scans.
Researchers at Stanford University have developed an artificial intelligence (AI) tool that can help detect brain aneurysms. The tool works by pinpointing areas of a brain scan that may contain an aneurysm.
RELATED: GOOGLE'S MEDICAL AI DETECTS LUNG CANCER WITH 94% ACCURACY
“There’s been a lot of concern about how machine learning will actually work within the medical field,” said Allison Park, a Stanford graduate student in statistics and co-lead author of the paper. “This research is an example of how humans stay involved in the diagnostic process, aided by an artificial intelligence tool.”
The tool helped clinicians correctly identify up to six more aneurysms in 100 scans. However, the team of researchers advises that further investigation is needed to evaluate generalizability of the AI tool before it can be released in real-time clinical applications. But still, the tool is a welcome addition as search for aneurysms is painfully difficult work.
After the training, the algorithm could decide for each voxel of a scan whether there is an aneurysm there. Better yet, the result of the AI tool comes up as the algorithm’s conclusions overlaid as a semi-transparent highlight on top of the scan, allowing the clinicians to still see what the scans look like without the AI’s input.
“We were interested how these scans with AI-added overlays would improve the performance of clinicians,” said Pranav Rajpurkar, a graduate student in computer science and co-lead author of the paper. “Rather than just having the algorithm say that a scan contained an aneurysm, we were able to bring the exact locations of the aneurysms to the clinician’s attention.”
Eight clinicians tested the novel tool by evaluating a set of 115 brain scans. With the tool, the clinicians correctly identified more aneurysms and were more likely to agree with one another on the final diagnosis.
Not designed to work with AI
The tool, believe the researchers, could now be further trained to identify other diseases. But an issue remains with this line of work. Current scan viewers and other machines are simply not designed to work with deep learning technology.
“Because of these issues, I think deployment will come faster not with pure AI automation, but instead with AI and radiologists collaborating,” said Andrew Ng, adjunct professor of computer science and co-senior author of the paper. “We still have technical and non-technical work to do, but we as a community will get there and AI-radiologist collaboration is the most promising path.”
The paper is published in June 7 in JAMA Network Open,
Comments
Post a Comment