Pragmatic Evaluation of a Deep-Learning Algorithm to Automate Ejection Fraction on Hand-Held, Point-of-Care Echocardiography in a Cardiac Surgical Operating Room

Abstract

ObjectiveTest the correlation of ejection fraction (EF) estimated by a deep learning-based, automated algorithm (Auto EF) vs an EF estimated by Simpson's method. DesignProspective observational study. SettingSingle-center study at the Hospital of the University of Pennsylvania. ParticipantsStudy participants were ≥18 years of age, scheduled to undergo valve, aortic, coronary artery bypass graft (CABG), heart or lung transplant surgery. InterventionsThis non-interventional study involved acquiring apical 4 Chamber (A4C), transthoracic echocardiographic clips using Philips hand-held ultrasound device; Lumify. Measurements and Main ResultsIn the primary analysis of 54 clips, compared to the Simpson's method for EF estimation, bias was similar for Auto EF (-10.17%) and the experienced reader-estimated EF (-9.82%), but correlation was lower for Auto EF (r = 0.56) than the experienced reader-estimated EF (r = 0.80). In the secondary analyses, correlation between EF estimated by Simpson's method and Auto EF increased when applied to 27 acquisitions classified as adequate (r = 0.86) but decreased when applied to 27 acquisitions classified as inadequate (r = 0.46). ConclusionsApplied to acquisitions of adequate image quality, Auto EF produced a numerical EF estimate equivalent to Simpson's method. However, when applied to acquisitions of inadequate image quality, discrepancies arose between EF estimated by Auto EF and Simpson's method. Visual EF estimates by experienced readers correlated highly with Simpson's method in both variable and inadequate imaging conditions – emphasizing its enduring clinical utility. Trial RegistrationClinicaltrials.gov trial identifier: NCT04943965