Could AI Reinvent Efficiency, Detection and Triage in Cardiac Screening with Handheld Ultrasound?

Could emerging ultrasound-based AI software facilitate improved access to cardiac function screening and earlier detection of significant cardiac dysfunction by clinicians who are not cardiologists?

For a recent study, published in Frontiers in Digital Health, researchers employed retrospective data from 120,127 echocardiographic studies to develop and validate a deep learning model (AISAP CARDIO V1.0, AISAP) for detection of right ventricular dysfunction, reduced left ventricular ejection fraction (LVEF) and moderate or greater mitral or tricuspid regurgitation.

After garnering high AUCs in retrospective testing, including 88.3 percent for mitral regurgitation and 98.2 percent for reduced LVEF, the study authors proceeded to explore the utility of the AI software on handheld ultrasound exams performed by non-cardiologists in a prospective trial of 209 patients.

In the prospective study, the study authors found that the AI software facilitated AUCs of 72 percent for mitral regurgitation, 87 percent for tricuspid regurgitation, 95 percent for right ventricular dysfunction and 97 percent for reduced ejection LVEF.

In a recent interview with Diagnostic Imaging, lead study author Lior Fisher, MD, discussed the advantages of the AI software in facilitating an efficient “abundancy of data” in cardiac function screening with only a single two-dimensional apical four-chamber echocardiographic view.

“The first thing is basically having a private tutor on point-of-care ultrasound with you all the time. The second thing is the speed. You do the test (and) you hit generate. It takes about 10 seconds. (When) you get back to your desk, it’s already done, and it gives you a report. It is very similar to the one that you would get from the echo core lab, but only in 50 seconds, not two days,” maintained Dr. Fisher, a physician at the Leviev Cardiovascular Institute and Sheba Medical Center in Ramat Gan, Israel.

In addition to possibly facilitating more timely triage in emergency departments, Dr. Fisher emphasized the potential of the AI software with handheld ultrasound devices to improve health equity in cardiac function screening.

“I think that maybe out of the hospital, have this device — which is very space-efficient and not as expensive — in your small town clinic to give a lot of people that don’t have access to major medical centers the opportunity and the right to have early diagnosis and to understand better if they have structural heart disease, especially the elderly population, underprivileged people and (those in) rural places. This is something that I think is very powerful for this technology,” added Dr. Fisher.

(Editor’s note: For related content, see “FDA Issues Expanded Coverage for UltraSight’s Cardiac Ultrasound Echo Stewardship Program,” “FDA Grants Fast Track Designation to Emerging Ultrasound Contrast Agent” and “GE HealthCare Launches Vivid Pioneer Ultrasound Platform at ESC Conference.”)