The AI, skilled in London, can detect COVID with out errors within the lung ultrasound

0
364

Breadcrumb Trail Links

A new artificial intelligence tool that interprets lung ultrasound can accurately identify COVID-19, London researchers say.

Author of the article:

Jennifer Bieman

Release date:

March 22, 2021 • • 17 hours ago • • Read for 2 minutes • • 6 comments The Lawson Health Research Institute team, Blake VanBerlo, Dr.  Robert Arntfield and Derek Wu, developed a program that can accurately determine the difference between pulmonary ultrasound from COVID-19 patients and those with non-coronavirus pneumonia.  (Post / Lawson Health Research Institute)The Lawson Health Research Institute team, Blake VanBerlo, Dr. Robert Arntfield and Derek Wu, developed a program that can accurately determine the difference between pulmonary ultrasound from COVID-19 patients and those with non-coronavirus pneumonia. (Post / Lawson Health Research Institute)

Article content

A new artificial intelligence tool that interprets lung ultrasound can accurately identify COVID-19, London researchers say.

Scientists at the Lawson Health Research Institute, the research arm of London’s hospitals, trained a computer to find and recognize different patterns in the lung ultrasound images of COVID-19 patients. These are hidden features that even trained specialists cannot detect.

The research team then tested the program to see if it could tell the difference between COVID-19 lungs and those with non-coronavirus pneumonia.

“We were a bit shocked. There was a perfect record of the separation of COVID from non-COVID pneumonia (ultrasound), ”said Robert Arntfield, a Lawson researcher and medical director of the London Health Sciences Centre’s Critical Care Trauma Center.

“I’m ready to acknowledge that 100 percent seems too good to be true, but in the science we’ve done and the way we’ve done it, it’s strong enough that we’re very confident that there is a signal like a QR code buried in the lungs that we cannot see, but that computer vision can extract. “

advertising

This ad hasn’t loaded yet, but your article continues below.

Article content

COVID-19 can cause serious, sometimes fatal, lung complications. Lung ultrasound exams of COVID-19 patients and patients with non-coronavirus pneumonia look similar, even to the trained eye, Arntfield said.

In the study, researchers showed 61 lung ultrasound-trained acute care physicians 25 unidentified ultrasound clips from patients with COVID-19 and non-coronavirus lung diseases. You have been asked to distinguish the COVID-19 ultrasound scans from the non-COVID ultrasound machines.

“They were no better than a coin toss sorting between them. They are highly qualified doctors, ”said Arntfield.

However, the complex computer processing system developed by the London researchers was able to tell the difference.

The artificial intelligence model was created using a dataset of lung ultrasound from 243 patients, including 612 video loops and nearly 121,400 images.

Ultrasound machines are portable, relatively common, and cheaper to use than other diagnostic imaging machines, including CT scans. A finely tuned artificial intelligence program that can quickly analyze ultrasound images of the lungs and help doctors diagnose would be a cornerstone, Arntfield said.

The technology could be used to track and treat respiratory symptoms of heart failure, he said. Analyzing lung ultrasound with the artificial intelligence program could also help doctors look for healthy lungs.

“Before the pandemic, many of us at home wonder if our cold is just a cold or if it’s in our lungs or if it’s pneumonia,” Arntfield said.

advertising

This ad hasn’t loaded yet, but your article continues below.

Article content

“The mere claim that a lung is normal is of tremendous value, and we are on track to do so next with our research empire which we are slowly building.”

The artificial intelligence analysis program builds on a subject in London, Arntfield said.

Over the past decade, London’s hospitals have amassed hundreds of thousands of hours of lung ultrasound video, resulting in millions of still images, he said. The team is continuing its research in London’s hospitals through the Deep Breathe project.

“London could have the leg in bigger centers to really look at lung ultrasound and all these other permutations. We put the team together, we have the data and we are certainly really inspired by these findings to go all-in, “said Arntfield.

The study was published in the medical journal BMJ Open.

[email protected]

Share this article on your social network

advertising

This ad hasn’t loaded yet, but your article continues below.

advertising

This ad hasn’t loaded yet, but your article continues below.

LFP Noon Round Up Newsletter

By clicking the “Subscribe” button, you agree to receive the above-mentioned newsletter from Postmedia Network Inc. You can unsubscribe at any time by clicking the unsubscribe link at the bottom of our emails. Postmedia Network Inc. | 365 Bloor Street East, Toronto, Ontario, M4W 3L4 | 416-383-2300

Remarks

Postmedia strives to maintain a lively but civil discussion forum and to encourage all readers to share their views on our articles. It can take up to an hour for comments to be moderated before they appear on the website. We ask that you keep your comments relevant and respectful. We have turned on email notifications. You will now receive an email when you get a reply to your comment, when a comment thread you’re following is updated, or when a user follows comments. For more information and details on customizing your email settings, see our Community Guidelines.