June 19, 2019 — DiA Imaging Analysis announced the presentation of two studies assessing the performance and accuracy of the company's LVivo EF artificial intelligence (AI)-based solution that generates fully automated left ventricular ejection fraction (LVEF). EF is a main indicator in assessing the global function of the left ventricle. Presently, many readers quantify EF visually or trace endocardial borders that may not be well defined, yielding high variability among results. LVivo EF aims to substantially reduce this variability by generating objective, accurate and fast AI-based EF.
The studies were led by Martin Goldman, M.D., Solomon Bienstock, M.D., Rajeev Samtani, M.D., Steve Liao, M.D., Usman Baber, M.D., Arieh Greenbaum, M.D., Lori B. Croft, M.D., and Eric Stern, M.D., from Mount Sinai Hospital, New York.
The first study, "Validation of a Novel Artificial Intelligence Left Ventricular Ejection Fraction Quantification Software (LVivo EF by DIA) by Cardiac MRI," tested the accuracy of the LVivo EF tool on 76 patients by retrospectively running it on cardiac four-chamber clips. The LVivo EF results were then compared to those obtained by magnetic resonance imaging (MRI) EF as a gold standard. The study indicated a strong correlation between cardiac MRI and LVivo EF (R2=0.89), and it suggests that LVivo EF may be used over a wide range of cardiac function, to evaluate EF in a fast and accurate manner.
"This study further validated DiA's AI-based tool, LVivo EF, against cardiac MRI EF, a recognized gold standard for the assessment of cardiac function. Based on the study's findings, LVivo EF has the potential to provide accurate and objective quantification of LVEF to support clinicians in the decision-making process, right at the patient's bedside, saving time and resources. This is specifically important in patients with low EF, where accuracy has clinical and therapeutic implications. Moreover, by providing the endocardial border overlay on the moving images, it also facilitates immediate confirmation of its accuracy by the reader," said Goldman, associate director of the CV Institute, director of the Fellowship Training Program and director of medical education at the Icahn School of Medicine at Mount Sinai Hospital.
The successful validation of LVivo EF as compared to MRI EF, led the team to publish a second study comparing LVivo EF results to physician's evaluation of EF using transthoracic echocardiography (TTE), with and without contrast agents, entitled "Fully Automated Echocardiographic Artificial Intelligence Software (LVivo EF by DiA) Could Replace Contrast Agents for Improving Accuracy of Left Ventricular Ejection Fraction Quantification." Results indicated that in non-contrast studies compared to cardiac MRI, LVivo EF was significantly better than physicians' assessments (R2=0.823 compared to R2=0.622), while for contrast studies, which are often used to improve LVEF analysis, LVivo EF on non-contrast images and physicians' quantification of contrast-enhanced images were similar (R2=0.913 and R2=0.873).
Both studies will be presented by Bienstock at the American Society of Echocardiography (ASE) 30th Annual Scientific Session, June 22-24 in Portland, Ore.
For more information: www.dia-analysis.com