EchoNet-Dynamic is a new deep learning algorithm developed using 10,030 echocardiogram videos to estimate left ventricular ejection fraction (LVEF) and classify patients with heart failure with similar accuracy to that of experienced cardiologists. Human interpretation of echocardiograms relies on segmenting the left ventricle over a small number of cardiac cycles, which can have high interobserver variability. The artificial intelligence (AI)-based algorithm incorporated information across multiple cardiac cycles and accurately segmented the left ventricle (Dice similarity coefficient 0.92), predicted LVEF (mean absolute error 4.1%) and classified heart failure with reduced ejection fraction (area under the curve 0.97). This performance was validated using an independent dataset and was more reproducible than that of clinicians.
References
Original article
Ouyang, D. et al. Video-based AI for beat-to-beat assessment of cardiac function. Nature https://doi.org/10.1038/s41586-020-2145-8 (2020)
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Lim, G.B. Estimating ejection fraction by video-based AI. Nat Rev Cardiol 17, 320 (2020). https://doi.org/10.1038/s41569-020-0375-y
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41569-020-0375-y