‘Groundbreaking study’: AI helps providers with zero experience capture high-quality heart ultrasounds

Using a new deep learning tool, providers with zero ultrasound experience captured echocardiography images capable of evaluating a handful of key heart-related metrics, according to a multicenter study published Thursday.

Expert cardiologists reviewed images gathered by eight nurses who used Caption Health’s AI platform to analyze patients. The scans were clear and accurate enough to assess left ventricular size and function, right ventricular size, and the presence of nontrivial pericardial effusion, researchers explained in JAMA Cardiology.

The investigators believe their results represent a leap forward for AI in medical imaging that may bring ultrasonography access to those currently without it.

"Cardiac ultrasound is a powerful diagnostic technology that can be helpful and even life-saving in many clinical settings,” Chris Moore, MD, an emergency medicine physician at Yale, who isn’t listed on the study, said in a statement. “However, its application may be limited by the availability of experienced users. This groundbreaking study represents a significant step forward in making this technology more widely accessible to patients wherever needed."

For the prospective study, eight nurses who had never performed an echocardiogram were trained to use the algorithm and asked to each scan 30 patients who were already set to undergo a cardiac ultrasound. Every participant was at least 18 and treated at Northwestern Memorial Hospital or the Minneapolis Heart Institute between March and May 2019.

After five trained echocardiographers compared the nurses’ scans to those performed by sonographers who didn't use AI, 92.5% of the images were in agreement. Additionally, AI-directed exams were deemed accurate enough to assess right ventricular size and function and the presence of pericardial effusion in 98.8% of patients and right ventricular size and function in 92.5%.

These results, the authors noted, were true regardless of patients’ gender, race and body mass index.

According to Caption Health, this study was the basis for the U.S. Food & Drug Administration’s decision to grant De Novo pathway status to the Caption AI platform back in February 2020.

"The study's remarkable agreement between nurses' scans and sonographers' scans shows that the use of AI like Caption Guidance could fundamentally change how we use medical imaging," first author Akhil Narang, MD, a cardiologist at Northwestern Medicine, said Thursday. "This will extend the abilities of healthcare providers to evaluate for different pathologies in critical care, emergency departments and other settings—and perhaps identify them even earlier with the assistance of AI."

Around the web

Its testing slate will kick off with the diagnostic and interventional radiology subspecialty exams in nuclear imaging on Jan. 10. 

EZaccessMD offers teleconsults with board-certified physicians coupled with mobile diagnostics, hoping to help health plan holders limit costs on facility-based visits. 

A new survey of ortho experts offers key clues to help improve musculoskeletal radiology services and interdisciplinary collaboration. 

Psychology researchers have used machine learning to wring useful two-year dementia trajectory predictions from more than 500 potentially contributing risk factors.

Trimed Popup