12-Lead ECG Reconstruction via Koopman Operators

32% of all global deaths in the world are caused by cardiovascular diseases. Early detection, especially for patients with ischemia or cardiac arrhythmia, is crucial. To reduce the time between symptoms onset and treatment, wearable ECG sensors were developed to allow for the recording of the full 12-lead ECG signal at home. However, if even a single lead is not correctly positioned on the body that lead becomes corrupted, making automatic diagnosis on the basis of the full signal impossible. In this work, we present a methodology to reconstruct missing or noisy leads using the theory of Koopman Operators. Given a dataset consisting of full 12-lead ECGs, we learn a dynamical system describing the evolution of the 12 individual signals together in time. The Koopman theory indicates that there exists a high-dimensional embedding space in which the operator which propagates from one time instant to the next is linear. We therefore learn both the mapping to this embedding space, as well as the corresponding linear operator. Armed with this representation, we are able to impute missing leads by solving a least squares system in the embedding space, which can be achieved efficiently due to the sparse structure of the system. We perform an empirical evaluation using 12-lead ECG signals from thousands of patients, and show that we are able to reconstruct the signals in such way that enables accurate clinical diagnosis. Equal contribution Technion Israel Institute of Technology, Haifa, Israel Google Research Shamir Medical Center, Zerifin, Israel and Sackler School of Medicine, Tel-Aviv University, Israel. Correspondence to: Tomer Golany <tomer.golany@cs.technion.ac.il>, Kira Radinsky <kirar@cs.technion.ac.il>, Daniel Freedman <danielfreedman@google.com>, Saar Minha <minha.saar@gmail.com>. Proceedings of the 38 th International Conference on Machine Learning, PMLR 139, 2021. Copyright 2021 by the author(s).

[1]  S. Steinhubl,et al.  Effect of a Home-Based Wearable Continuous ECG Monitoring Patch on Detection of Undiagnosed Atrial Fibrillation: The mSToPS Randomized Clinical Trial , 2018, JAMA.

[2]  Mohammad Hossein Khosravi,et al.  Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980–2017: a systematic analysis for the Global Burden of Disease Study 2017 , 2018, Lancet.

[3]  Andrzej Banaszuk,et al.  Comparison of systems with complex behavior , 2004 .

[4]  Naoya Takeishi,et al.  Learning Koopman Invariant Subspaces for Dynamic Mode Decomposition , 2017, NIPS.

[5]  Qingxue Zhang,et al.  All-ECG: A Least-number of Leads ECG Monitor for Standard 12-lead ECG Tracking during Motion* , 2019, 2019 IEEE Healthcare Innovations and Point of Care Technologies, (HI-POCT).

[6]  I. Mezić,et al.  Analysis of Fluid Flows via Spectral Properties of the Koopman Operator , 2013 .

[7]  E. Frank.  An Accurate, Clinically Practical System For Spatial Vectorcardiography , 1956, Circulation.

[8]  Thomas B. Schön,et al.  Automatic diagnosis of the 12-lead ECG using a deep neural network , 2020, Nature Communications.

[9]  Rickey E. Carter,et al.  Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram , 2019, Nature Medicine.

[10]  Kira Radinsky,et al.  SimGANs: Simulator-Based Generative Adversarial Networks for ECG Synthesis to Improve Deep ECG Classification , 2020, ICML.

[11]  Kenneth Levenberg.  A METHOD FOR THE SOLUTION OF CERTAIN NON – LINEAR PROBLEMS IN LEAST SQUARES , 1944 .

[12]  J. Dimarco,et al.  Use of ambulatory electrocardiographic (Holter) monitoring. , 1990, Annals of internal medicine.

[13]  I. Mezić,et al.  Spectral analysis of nonlinear flows , 2009, Journal of Fluid Mechanics.

[14]  B. O. Koopman,et al.  Hamiltonian Systems and Transformation in Hilbert Space. , 1931, Proceedings of the National Academy of Sciences of the United States of America.

[15]  S. Brunton,et al.  Discovering governing equations from data by sparse identification of nonlinear dynamical systems , 2015, Proceedings of the National Academy of Sciences.

[16]  I. Mezić.  Spectral Properties of Dynamical Systems, Model Reduction and Decompositions , 2005 .

[17]  B. O. Koopman,et al.  Dynamical Systems of Continuous Spectra. , 1932, Proceedings of the National Academy of Sciences of the United States of America.

[18]  P. Laguna,et al.  New algorithm for QT interval analysis in 24-hour Holter ECG: performance and applications , 2006, Medical and Biological Engineering and Computing.

[19]  Michael W. Mahoney,et al.  Physics-informed Autoencoders for Lyapunov-stable Fluid Flow Prediction , 2019, ArXiv.

[20]  Soumya Kundu,et al.  Learning Deep Neural Network Representations for Koopman Operators of Nonlinear Dynamical Systems , 2017, 2019 American Control Conference (ACC).

[21]  Na Liu,et al.  A novel method based on convolutional neural networks for deriving standard 12-lead ECG from serial 3-lead ECG , 2019, Frontiers of Information Technology & Electronic Engineering.

[22]  J. Saul,et al.  Assessment of the 12-lead ECG as a screening test for detection of cardiovascular disease in healthy general populations of young people (12-25 Years of Age): a scientific statement from the American Heart Association and the American College of Cardiology. , 2014, Circulation.

[23]  Frank Noé,et al.  Time-lagged autoencoders: Deep learning of slow collective variables for molecular kinetics , 2017, The Journal of chemical physics.

[24]  Patrick E. McSharry,et al.  A dynamical model for generating synthetic electrocardiogram signals , 2003, IEEE Transactions on Biomedical Engineering.

[25]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[26]  P. Rubel,et al.  A neural network approach for patient-specific 12-lead ECG synthesis in patient monitoring environments , 2004, Computers in Cardiology, 2004.

[27]  Karthik Duraisamy,et al.  Physics-Informed Probabilistic Learning of Linear Embeddings of Nonlinear Dynamics with Guaranteed Stability , 2019, SIAM J. Appl. Dyn. Syst..

[28]  P. Schmid,et al.  Dynamic mode decomposition of numerical and experimental data , 2008, Journal of Fluid Mechanics.

[29]  V. Somers,et al.  Heart Rate Variability: , 2003, Journal of cardiovascular electrophysiology.

[30]  I. Mezić,et al.  Applied Koopmanism. , 2012, Chaos.

[31]  Steven L. Brunton,et al.  Deep learning for universal linear embeddings of nonlinear dynamics , 2017, Nature Communications.

[32]  Hao Wu,et al.  VAMPnets for deep learning of molecular kinetics , 2017, Nature Communications.

[33]  J A Scherer,et al.  Synthesis of the 12-lead electrocardiogram from a 3-lead subset using patient-specific transformation vectors. An algorithmic approach to computerized signal synthesis. , 1989, Journal of electrocardiology.

[34]  Daniel Freedman,et al.  ECG ODE-GAN: Learning Ordinary Differential Equations of ECG Dynamics via Generative Adversarial Learning , 2021, AAAI.

[35]  Ali Bahrami Rad,et al.  Classification of 12-lead ECGs: the PhysioNet/Computing in Cardiology Challenge 2020 , 2020, 2020 Computing in Cardiology.