por
John R. Fischer, Senior Reporter | April 11, 2022
From the April 2022 issue of HealthCare Business News magazine
But while POCUS has changed the landscape of cardiac ultrasound over the last 10 years, the technology still has drawbacks, according to Michael Blaivas, vice chair of AIUM Artificial Intelligence Community of Practices and a professor of medicine, affiliate at the University of South Carolina School of Medicine. “Training is typically required to acquire and interpret images. Even if you’re going to save them for somebody else to review, you still need to know what you’re doing to get quality images. The other challenge is having handheld or portable imaging devices that deliver images that are of adequate quality to be interpreted. We’ve only started to see that in the last few years.”
Automating speed and accuracy
In medical imaging, the development of AI is geared toward faster acquisition and more accurate interpretation of scans. One area in echocardiography where AI has helped is the interpretation of left ventricle size and function. In addition, there are now discussions about using pattern recognition to assess heart valves, according to Goldman.
He adds, however, that while exciting, AI for heart assessment is still a work-in-progress. “I think that artificial intelligence in echocardiography has a lot of promise and may be ready for prime time in certain settings. There are so many aspects of an echocardiogram. It's a dynamic study with potentially many views and potentially 50 or more 2-3 heartbeat-long loops of moving heart within a breathing patient, which makes it very challenging.”
Many companies are working to address this by designing applications that can select appropriate loops on which to perform AI and provide quantitative information on left and right heart function, size and even strain. “That's very exciting, but it still has to be vetted and proved to be working well,” said Goldman.
AI is also used to improve the quality of scans in various ways, says Lee. “There are a lot of different technologies to smooth the image out, offer better spatial resolution that create clear and crisper images. We’re also seeing a big increase in 3D and 4D rendering and 3D and 4D usage in evaluating the heart and other organs in the body. You can take a rendered image and evaluate that image in different planes, the ability to rotate the image aids in diagnosis and improved patient care.”
AI also can be used for auto calculation and offer doppler capabilities for examining heart function. But producing and interpreting quality images accurately still comes down to how well the sonographer performs a scan. “We are seeing applications that guide the user on how to scan, that say ‘rotate the probe this way, angle the probe that way. Now you have a good view. Nope, now you have a bad view, change the following.’ This is how you lower the skill level required for the cardiac ultrasound operator and you only have to look down the horizon so far before you see that eventually patients will be able to do this at home themselves because the device will be able to instruct the patients themselves,” said Blaivas.
