Estimation of left ventricular function using a novel acoustic-based device

Abstract

We examined the feasibility of estimating left ventricular ejection fraction (LVEF) by a novel acoustic-based device [vibration response imaging (VRI); Deep Breeze]. One hundred and forty-one subjects (117 patients and 24 healthy volunteers; age 55 ± 15 years, 82% men) were examined by both VRI and echocardiography. LVEF was determined by echocardiography (echo-LVEF) using the biplane Simpson's method. Low-frequency acoustic signals (10-70 Hz) were recorded by VRI from the left posterior thorax by a matrix of 36 microphones during 8 s of breath holding, and an electrocardiogram was recorded simultaneously. The acoustic signals were processed digitally, and an algorithm designed to estimate LVEF was developed (VRI-LVEF), based on a combination of multiple acoustic (systolic and diastolic acoustic signals, beat-to-beat variability of acoustic signals and propagation of acoustic signals throughout the matrix), electrocardiographic and clinical parameters. Mean echo-LVEF was 51 ± 15% (range, 11-76%). Echo-LVEF was reduced (< 50%) in 55 subjects (39%) and severely reduced (< 35%) in 28 subjects (20%). VRI-LVEF calculated by a multivariate algorithm correlated significantly with echo-LVEF (R(2) = 0·59; P < 0·001). VRI-LVEF accurately predicted the presence of reduced (< 50%) or severely reduced (< 35%) echo-LVEF, with sensitivities of 84% and 82%, specificities of 86% and 91%, positive predictive values of 79% and 70% and negative predictive values of 89% and 95%, respectively. LVEF can be estimated using a novel acoustic-based device. This device may assist in triage of patients according to LVEF prior to definitive assessment of LVEF by echocardiography.