Deep learning model for ECG reconstruction reveals the information content of ECG leads

• Autorzy: Gradowski Tomasz , Buchner Teodor

Identyfikatory

DOI

10.5604/01.3001.0055.1381

• Języki publikacji: EN

Abstrakty

EN

Objective: This study aimed to evaluate the information content of individual electrocardiogram (ECG) leads and their inter-lead correlations using a deep learning approach. Specifically, we investigated the capability of a neural network to reconstruct missing ECG leads from reduced-lead configurations, thereby revealing each lead's unique and shared informational value. Methods: We developed a U-net convolutional neural network to reconstruct missing leads in 12-lead ECG recordings. The model was trained using the PTB-XL dataset and tested on the PTB dataset. We trained the model with varying combinations of input leads, including single limb leads, combinations of two limb leads and configurations, including one or two precordial leads. We evaluated the model's performance using mean squared error (MSE) between the reconstructed and actual signals. Results: The models demonstrated varying reconstruction accuracy depending on the input lead configuration. Precordial leads V1 and V6 showed the highest reconstruction fidelity from limb leads alone, while V3 consistently exhibited the lowest, indicating its unique informational content. Conclusions: The proposed method effectively quantifies the informational value of ECG leads. This has significant implications for optimizing lead selection in diagnostic scenarios, particularly in settings where complete 12-lead ECGs are impractical. In addition, the study provides insights into the physiological underpinnings of ECG signals and their propagation. The findings pave the way for advances in telemedicine, portable ECG devices and personalized cardiac diagnostics by reducing redundancy and improving signal interpretation.

Słowa kluczowe

EN

electrocardiogram; ECG; lead reconstruction; convolutional neural network; CNN; U-Net; deep learning; machine learning

• Wydawca: Uniwersytet Jagielloński - Collegium Medicum ; Index Copernicus Sp. z o.o.
• Czasopismo: Bio-Algorithms and Med-Systems
• Rocznik: 2025
• Tom: Vol. 21, no 1
• Strony: 13--23
• Opis fizyczny: Bibliogr. 27 poz., rys.

Twórcy

autor

Gradowski Tomasz

• Faculty of Physics, Warsaw University of Technology, Warsaw, Poland

• https://orcid.org/0000-0001-5418-2071

autor

Buchner Teodor

• Faculty of Physics, Warsaw University of Technology, Warsaw, Poland

• https://orcid.org/0000-0003-0030-3194

Bibliografia

• 1. Yoo H, Yum Y, Kim Y, Kim JH, Park HJ, Joo HJ. Restoration of missing or low-quality 12-lead ECG signals using ensemble deep-learning model with optimal combination. Biomed. Signal Process. Control. 2023;83:104690. doi: https://doi.org/10.1016/j.bspc.2023.104690.
• 2. Xue J. Does a reduced ECG lead set contain the full 12-lead ECG information for interpretation?. CinC. 2024;51:1-4. doi: https://doi.org/10.22489/cinc.2024.473.
• 3. Pipberger HV, Bialek SM, Perloff JK, Schnaper HW. Correlation of clinical information in the standard 12-lead ECG and in a corrected orthogonal 3-lead ECG. Am. Heart J. 1961;61(1):34-43. doi: https://doi.org/10.1016/0002-8703(61)90514-2.
• 4. Holderith M, Schanze T. Cross-Correlation based comparison between the conventional 12-lead ECG and an EASI derived 12-lead ECG. Curr. Dir. Biomed. Eng. 2018;4(1):621-4. doi: https://doi.org/10.1515/cdbme-2018-0149.
• 5. Jain U, Butchy A, Leasure M, Covalesky V, McCormick D, Mintz G, editors. Redundancy and novelty between ECG leads based on linear correlation. Proceedings of the 16th international joint conference on biomedical engineering systems and technologies; 2023; Lisbon, Portugal. SCITEPRESS - Science; Technology Publications; 2023. pp. 359-365. doi: https://doi.org/10.5220/0011815700003414.
• 6. Zhang C, Li J, Pang S, Xu F, Zhou S. A 12-lead ECG correlation network model exploring the inter-lead relationships. Europhys Lett 2022;140(3):31001. doi: https://doi.org/10.1209/0295- 5075/ac9b89.
• 7. Lence A, Granese F, Fall A, Hanczar B, Salem J-E, Zucker J-D, et al. ECGrecover: A deep learning approach for electrocardiogram signal completion. KDD ‘25: Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1 2025;2359-70. doi: https://doi.org/10.1145/3690624.3709405.
• 8. Frank E. An Accurate, Clinically Practical System For Spatial Vectorcardiography. Circulation. 1956;13(5):737-49. doi: https://doi.org/10.1161/01.cir.13.5.737.
• 9. Gargiulo GD. True Unipolar ECG Machine for Wilson Central Terminal Measurements. BioMed Res Int. 2015;2015:586397. doi: https://doi.org/10.1155/2015/586397.
• 10. Gargiulo G, Bifulco P, Cesarelli M, McEwan A, O’Loughlin A, Tapson J, Thiagalingam A. The Wilson’s Central Terminal (WCT): A Systematic Error in ECG Recordings. Heart Lung Circ. 2016;25:S263. doi: https://doi.org/10.1016/j.hlc.2016.06.613.
• 11. Moeinzadeh H, Bifulco P, Cesarelli M, McEwan AL, O’Loughlin A, Shugman IM, et al. Minimization of the Wilson’s Central Terminal voltage potential via a genetic algorithm. BMC Res. Notes. 2018;11(1). doi: https://doi.org/10.1186/s13104-018-4017-y.
• 12. Man S, Maan AC, Schalij MJ, Swenne CA. Vectorcardiographic diagnostic & prognostic information derived from the 12‐lead electrocardiogram: Historical review and clinical perspective. J. Electrocardiol. 2015;48(4):463-75. doi: https://doi.org/10.1016/j.jelectrocard.2015.05.002.
• 13. Boonstra MJ, Brooks DH, Loh P, van Dam PM. Cine ECG: A novel method to image the average activation sequence in the heart from the 12-lead ECG. Comput Biol Med. 2022;141(105128):105128. doi: https://doi.org/10.1016/j.compbiomed.2021.105128.
• 14. Schmitt OH, Levine RB, Simonson E. Electrocardiographic mirror pattern studies. I. Am J. 1953;45(3):416-28. doi: https://doi.org/10.1016/0002-8703(53)90152-5.
• 15. Brody DA, Copeland GD. Electrocardiographic cancellation: Some observations concerning the „nondipolar” fraction of precordial electrocardiograms. Am Heart J. 1958;56(3):381-95. doi: https://doi.org/10.1016/0002-8703(58)90277-1.
• 16. Vaidya GN, Antoine S, Imam SH, Kozman H, Smulyan H, Villarreal D. Reciprocal ST-Segment Changes in Myocardial Infarction: Ischemia at Distance Versus Mirror Reflection of ST-Elevation. Am J Med Sci. 2018;355(2):162-7. doi: https://doi.org/10.1016/j.amjms.2017.09.004.
• 17. Wang J, Li J, Diao S, Xu H, Ding F. Atypical de Winter ECG pattern may be the mirror image of ST elevation. Ann Noninvasive Electrocardiol. 2022;27(3):e12915. doi: https://doi.org/10.1111/anec.12915.
• 18. Meek S. ABC of clinical electrocardiography: Introduction. II - Basic terminology. BMJ. 2002;324(7335):470-473. doi: https://doi.org/10.1136/bmj.324.7335.470.
• 19. Pérez-Riera AR, Barbosa-Barros R, Daminello-Raimundo R, de Abreu LC. Main artifacts in electrocardiography. Ann Electrocardiol. 2018;23(2):e12494. doi: https://doi.org/10.1111/anec.12494.
• 20. Buchner T, Zajdel M, Pęczalski K, Nowak P. Finite velocity of ECG signal propagation: preliminary theory, results of a pilot experiment and consequences for medical diagnosis. Sci Rep. 2023;13(1):4716. doi: https://doi.org/10.1038/s41598-023-29904-2.
• 21. Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation. In: Navab N, Hornegger J, Wells WM, Frangi AF, editors. Medical image computing and computer-assisted intervention - MICCAI 2015. Cham: Springer International Publishing; 2015. pp. 234-41. doi: https://doi.org/10.1007/978-3-319-24574-4\_28.
• 22. Goodfellow I, Bengio Y, Courville A. Deep learning. Adaptive computation and machine learning. Cambridge: Mass: The MIT press; 2016. pp. 499-523.
• 23. Wagner P, Strodthoff N, Bousseljot R, Samek W, Schaeffter T. PTBXL, a large publicly available electrocardiography dataset. PhysioNet. 2022;101(23):e215-e220. doi: https://doi.org/10.13026/KFZX-AW45.
• 24. Wagner P, Strodthoff N, Bousseljot R-D, Samek W, Schaeffter T. PTBXL, a large publicly available electrocardiography dataset. Sci. Data. 2020;7(1):154.
• 25. Bousseljot R, Kreiseler D, Schnabel A. Nutzung der EKG-signaldatenbank CARDIODAT der PTB über das internet. Biomedizinische Technik/ Biomedical Engineering. 2009;40:317-8. doi: https://doi.org/10.1515/ bmte.1995.40.s1.317.
• 26. Goldberger AL, Amaral LA, Glass L, Hausdorff JM, Ivanov PC, Mark RG, et al. PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation. 2000;101(23):E215-20.
• 27. Siontis KC, Noseworthy PA, Attia ZI, Friedman PA. Artificial intelligence-enhanced electrocardiography in cardiovascular disease management. Nat Rev Cardiol. 2021;18(7):465-78. doi: https://doi.org/10.1038/s41569-020-00503-2.