Be part of the knowledge.

We’re glad to see you’re enjoying ReachMD…
but how about a more personalized experience?

Register for free

New AI Tool Helps Radiologists Detect Brain Aneurysms

New AI Tool Helps Radiologists Detect Brain Aneurysms

Artificial intelligence can now detect brain aneurysms.

Stanford University researchers developed an AI tool to highlight areas of brain scans likely to contain an aneurysm.

The condition—which can go unnoticed until it’s too late—causes blood vessels in the brain to bulge to the point of leaking or bursting, potentially causing stroke, brain injury, or death.

“There’s been a lot of concern about how machine learning will actually work within the medical field,” Allison Park, a Stanford graduate student and co-lead author of the study, said in a statement. “This research is an example of how humans stay involved in the diagnostic process, aided by an artificial intelligence tool.”

Aneurysms come in various shapes and sizes: some balloon out at tricky angles, others register as no more than a blip. It can take humans hours to comb through scans for signs of irregularities.

“Search[ing] for an aneurysm is one of the most labor-intensive and critical tasks radiologists undertake,” according to Kristen Yeom, associate professor of radiology and co-senior author of the paper.

“Given inherent challenges of complex neurovascular anatomy and potential fatal outcome of a missed aneurysm,” she continued, “it prompted me to apply advances in computer science and vision to neuroimaging.”

So Yeom recruited a team of AI experts—including Stanford’s Machine Learning Group leader Andrew Ng and computer science graduate student Christopher Chute—to create an artificial intelligence tool that can accurately process large stacks of 3D images and complement clinical diagnostic practice.

Which is easier said than done.

To cultivate their algorithm, the team labeled, by hand, every voxel (the 3D equivalent to a pixel) based on whether or not it was part of an aneurysm.

“Building the training data was a pretty grueling task and there were a lot of data,” Chute said.

The HeadXNet tool’s conclusions are then overlaid as a semi-transparent highlight on top of the scan, allowing clinicians to still see what the original scan looks like without the AI’s input.

“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,” Pranav Rajpurkar, a graduate student in computer science, said.

Eight doctors tested HeadXNet by evaluating a set of 115 brain scans—once with the help of the AI and once without.

With the tool, clinicians not only correctly identified more aneurysms, but they were also more likely to agree with one another.

According to the findings, HeadXNet did not influence how long it took the physician to decide on a diagnosis or their ability to correctly identify scans without aneurysms—”a guard against telling someone they have an aneurysm when they don’t.”

That’s great news for doctors and patients alike.

But despite the early success of HeadXNet, there is still plenty of work to be done.

Researchers caution that further investigation is needed before the system can roll out—mainly due to differences in scanner hardware and imaging protocols across hospital centers.

“I think deployment will come faster not with pure AI automation, but instead with AI and radiologists collaborating,” Ng said. “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.”

In the future, the machine learning methods at the heart of this program could be trained to identify other diseases inside and outside the brain.

Facebook Comments

LIVE ON REACHMD RADIOBack to live radio


Programs 8/25/19

Get a Dose of ReachMD in Your Inbox
and Practice Smarter Medicine

Stay current with the best in medical education.