{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T21:57:25Z","timestamp":1790287045508,"version":"4.1.0"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T00:00:00Z","timestamp":1768176000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Physiol."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Non-invasive estimation of intracardiac electrograms (EGMs) from body surface potential measurements (BSPMs) could reduce reliance on invasive mapping and enable safer, patient-specific characterization of atrial arrhythmias. Conventional inverse problem formulations, such as Tikhonov regularization, are limited by ill-posedness, sensitivity to anatomical inaccuracies, and low spatial resolution.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>In this work, we propose a dual-branch deep learning (DL) architecture based on a variational autoencoder (VAE) to directly reconstruct atrial EGMs from BSPMs.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>A dataset of 680 BSPM-EGM pairs was generated using biatrial computational models simulating a wide spectrum of rhythms, including sinus rhythm, atrial fibrillation (AF), ectopic activity, and fibrotic substrates. The network learns a shared latent representation of BSPMs, simultaneously optimized for BSPM self-reconstruction and EGM prediction. Performance was assessed across two phases: a baseline dataset with well-represented rhythms (sinus and multirotor AF), and an extended dataset with rarer arrhythmic classes. Evaluation employed multiple temporal and spectral metrics, as well as spatial voltage and phase mapping.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Results show that stratified training yielded the most balanced performance, particularly in AF, with improved correlation, peak detection precision, and spectral coherence compared to baseline and regularized models. Against the zero-order Tikhonov method, the proposed DL model preserved waveform morphology and spectral content more faithfully across rhythm types.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>These findings demonstrate that non-invasive, data-driven EGM reconstruction is feasible and can capture physiologically relevant temporal and spatial dynamics. By providing more coherent functional information from BSPMs, DL-based approaches may support individualized diagnosis and guide ablation strategies in atrial arrhythmia care.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fphys.2025.1720244","type":"journal-article","created":{"date-parts":[[2026,1,12]],"date-time":"2026-01-12T08:13:02Z","timestamp":1768205582000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["Deep learning for atrial electrogram estimation: toward non-invasive arrhythmia mapping using variational autoencoders"],"prefix":"10.3389","volume":"16","author":[{"given":"Miriam","family":"Guti\u00e9rrez-Fern\u00e1ndez","sequence":"first","affiliation":[{"name":"Signal Theory and Communications Dpt., EIF, Universidad Rey Juan Carlos","place":["Fuenlabrada, Spain"]},{"name":"Vicomtech Foundation, Basque Research and Technology Alliance (BRTA)","place":["San Sebasti\u00e1n, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"K.","family":"L\u00f3pez-Linares","sequence":"additional","affiliation":[{"name":"Vicomtech Foundation, Basque Research and Technology Alliance (BRTA)","place":["San Sebasti\u00e1n, Spain"]},{"name":"eHealth Group, Bioengineering Area, Biogipuzkoa Health Research Institute","place":["San Sebasti\u00e1n, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"C.","family":"Fambuena-Santos","sequence":"additional","affiliation":[{"name":"ITACA Institute, Universitat Polit\u00e8cnica de Val\u00e8ncia","place":["Val\u00e8ncia, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maria S.","family":"Guillem","sequence":"additional","affiliation":[{"name":"ITACA Institute, Universitat Polit\u00e8cnica de Val\u00e8ncia","place":["Val\u00e8ncia, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andreu M.","family":"Climent","sequence":"additional","affiliation":[{"name":"ITACA Institute, Universitat Polit\u00e8cnica de Val\u00e8ncia","place":["Val\u00e8ncia, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"\u00d3.","family":"Barquero-P\u00e9rez","sequence":"additional","affiliation":[{"name":"Signal Theory and Communications Dpt., EIF, Universidad Rey Juan Carlos","place":["Fuenlabrada, Spain"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,1,12]]},"reference":[{"key":"B1","doi-asserted-by":"crossref","DOI":"10.1145\/3292500.3330701","article-title":"Optuna: a next-generation hyperparameter optimization framework","author":"Akiba","year":"2019"},{"key":"B2","doi-asserted-by":"publisher","first-page":"975","DOI":"10.1016\/j.jelectrocard.2015.08.008","article-title":"Experimental data and geometric analysis repository (edgar)","volume":"48","author":"Aras","year":"2015","journal-title":"J. Electrocardiol."},{"key":"B3","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1016\/j.hrthm.2008.10.024","article-title":"Real-time dominant frequency mapping and ablation of dominant frequency sites in atrial fibrillation with left-to-right frequency gradients predicts long-term maintenance of sinus rhythm","volume":"6","author":"Atienza","year":"2009","journal-title":"Heart Rhythm."},{"key":"B4","doi-asserted-by":"publisher","first-page":"euac053.566","DOI":"10.1093\/europace\/euac053.566","article-title":"Deep learning formulation of ECGI evaluated on clinical data","volume":"24","author":"Bacoyannis","year":"2022","journal-title":"EP Eur."},{"key":"B5","doi-asserted-by":"publisher","first-page":"1188","DOI":"10.1093\/europace\/eur112","article-title":"Direct visualization of the slow pathway using voltage gradient mapping: a novel approach for successful ablation of atrioventricular nodal reentry tachycardia","volume":"13","author":"Bailin","year":"2011","journal-title":"Europace"},{"key":"B6","volume-title":"Random data: analysis and measurement procedures","author":"Bendat","year":"2011"},{"key":"B7","doi-asserted-by":"publisher","first-page":"102354","DOI":"10.1016\/j.bspc.2020.102354","article-title":"Electrocardiographic imaging including intracardiac information to achieve accurate global mapping during atrial fibrillation","volume":"64","author":"C\u00e1mara-V\u00e1zquez","year":"","journal-title":"Biomed. Signal Process. Control"},{"key":"B8","doi-asserted-by":"publisher","first-page":"733449","DOI":"10.3389\/fphys.2021.733449","article-title":"Non-invasive estimation of atrial fibrillation driver position with convolutional neural networks and body surface potentials","volume":"12","author":"C\u00e1mara-V\u00e1zquez","year":"","journal-title":"Front. Physiology"},{"key":"B9","doi-asserted-by":"publisher","first-page":"2331","DOI":"10.3390\/s22062331","article-title":"Solving inverse electrocardiographic mapping using machine learning and deep learning frameworks","volume":"22","author":"Chen","year":"2022","journal-title":"Sensors"},{"key":"B10","doi-asserted-by":"publisher","first-page":"459","DOI":"10.1016\/j.ccep.2019.05.004","article-title":"Noninvasive mapping and electrocardiographic imaging in atrial and ventricular arrhythmias (cardioinsight)","volume":"11","author":"Cheniti","year":"2019","journal-title":"Card. Electrophysiology Clinics"},{"key":"B11","first-page":"395","article-title":"Image super-resolution with deep variational autoencoders","author":"Chira","year":"2022"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1364","DOI":"10.1161\/CIRCULATIONAHA.110.945709","article-title":"Noninvasive characterization of epicardial activation in humans with diverse atrial fibrillation patterns","volume":"122","author":"Cuculich","year":"2010","journal-title":"Circulation"},{"key":"B13","doi-asserted-by":"publisher","first-page":"404","DOI":"10.1038\/s41569-022-00820-8","article-title":"Epidemiology and modifiable risk factors for atrial fibrillation","volume":"20","author":"Elliott","year":"2023","journal-title":"Nat. Rev. Cardiol."},{"key":"B14","doi-asserted-by":"publisher","first-page":"466","DOI":"10.3389\/fphys.2016.00466","article-title":"Regularization techniques for ecg imaging during atrial fibrillation: a computational study","volume":"7","author":"Figuera","year":"2016","journal-title":"Front. Physiology"},{"key":"B15","article-title":"Solving bayesian inverse problems via variational autoencoders","author":"Goh","year":"2019"},{"key":"B16","doi-asserted-by":"publisher","first-page":"156","DOI":"10.1111\/jce.13149","article-title":"Substrate ablation for treatment of atrial fibrillation: back to basics","volume":"28","author":"Goldberger","year":"2017","journal-title":"J. Cardiovasc. Electrophysiol."},{"key":"B17","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1016\/S1131-3587(16)30007-3","article-title":"Epidemiolog\u00eda de la fibrilaci\u00f3n auricular","author":"G\u00f3mez-Doblasa","year":"2016","journal-title":"Rev. Esp. Cardiol. Supl."},{"key":"B18","doi-asserted-by":"publisher","first-page":"108052","DOI":"10.1016\/j.cmpb.2024.108052","article-title":"Non-invasive estimation of atrial fibrillation driver position using long-short term memory neural networks and body surface potentials","volume":"246","author":"Guti\u00e9rrez-Fern\u00e1ndez-Calvillo","year":"2024","journal-title":"Comput. Methods Programs Biomed."},{"key":"B19","first-page":"1","article-title":"Egm reconstruction from bsps in atrial fibrillation using deep learning","author":"Guti\u00e9rrez-Fern\u00e1ndez","year":"2023"},{"key":"B20","doi-asserted-by":"publisher","first-page":"530","DOI":"10.1161\/CIRCULATIONAHA.113.005421","article-title":"Driver domains in persistent atrial fibrillation","volume":"130","author":"Haissaguerre","year":"2014","journal-title":"Circulation"},{"key":"B21","doi-asserted-by":"publisher","first-page":"879","DOI":"10.1007\/s11517-022-02709-7","article-title":"Electrocardiographic imaging in the atria","volume":"61","author":"Hern\u00e1ndez-Romero","year":"2023","journal-title":"Med. and Biol. Eng. and Comput."},{"key":"B22","doi-asserted-by":"publisher","first-page":"87","DOI":"10.1117\/12.2559808","article-title":"Srvae: super resolution using variational autoencoders","volume":"11400","author":"Heydari","year":"2020","journal-title":"Pattern Recognit. and Track. XXXI"},{"key":"B23","doi-asserted-by":"publisher","first-page":"1735","DOI":"10.1162\/neco.1997.9.8.1735","article-title":"Long short-term memory","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural Comput."},{"key":"B24","doi-asserted-by":"publisher","first-page":"20991","DOI":"10.1007\/s11042-021-10600-0","article-title":"Sensor-data augmentation for human activity recognition with time-warping and data masking","volume":"80","author":"Jeong","year":"2021","journal-title":"Multimedia Tools Appl."},{"key":"B25","doi-asserted-by":"publisher","first-page":"iii28","DOI":"10.1093\/europace\/eun225","article-title":"Electro-anatomic mapping systems in arrhythmias","volume":"10","author":"Knackstedt","year":"2008","journal-title":"Europace"},{"key":"B26","first-page":"950","article-title":"A simple weight decay can improve generalization","volume":"4","author":"Krogh","year":"1992","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"B27","doi-asserted-by":"publisher","first-page":"341","DOI":"10.1111\/pace.14140","article-title":"Anatomical targets and expected outcomes of catheter-based ablation of atrial fibrillation in 2020","volume":"44","author":"La","year":"2021","journal-title":"Pacing Clin. Electrophysiol."},{"key":"B28","first-page":"1","article-title":"ECGI periodicity unraveled: a deep learning approach for the visualization of periodic spatiotemporal patterns in atrial fibrillation patients","author":"Lacki","year":"2021"},{"key":"B29","first-page":"177","article-title":"C1 surface interpolation for scattered data on a sphere","author":"Lawson","year":"1984","journal-title":"Rocky Mt. J. Math."},{"key":"B30","doi-asserted-by":"publisher","first-page":"S23","DOI":"10.1111\/j.1540-8167.2006.00630.x","article-title":"Substrate ablation in treatment of atrial fibrillation","volume":"17","author":"Pappone","year":"2006","journal-title":"J. Cardiovasc. Electrophysiol."},{"key":"B31","doi-asserted-by":"publisher","first-page":"240","DOI":"10.1098\/rspl.1895.0041","article-title":"Vii. note on regression and inheritance in the case of two parents","volume":"58","author":"Pearson","year":"1895","journal-title":"Proc. R. Soc. Lond."},{"key":"B32","doi-asserted-by":"publisher","first-page":"435","DOI":"10.1111\/jce.12931","article-title":"Noninvasive estimation of epicardial dominant high-frequency regions during atrial fibrillation","volume":"27","author":"Pedr\u00f3n-Torrecilla","year":"2016","journal-title":"J. Cardiovasc. Electrophysiol."},{"key":"B33","first-page":"22894","article-title":"Inverse problem regularization with hierarchical variational autoencoders","author":"Prost","year":"2023"},{"key":"B34","doi-asserted-by":"publisher","first-page":"e007700","DOI":"10.1161\/CIRCEP.119.007700","article-title":"Noninvasive assessment of complexity of atrial fibrillation: correlation with contact mapping and impact of ablation","volume":"13","author":"Rodrigo","year":"2020","journal-title":"Circulation Arrhythmia Electrophysiol."},{"key":"B35","doi-asserted-by":"publisher","first-page":"e005008","DOI":"10.1161\/CIRCEP.117.005008","article-title":"Technical considerations on phase mapping for identification of atrial reentrant activity in direct-and inverse-computed electrograms","volume":"10","author":"Rodrigo","year":"2017","journal-title":"Circulation Arrhythmia Electrophysiol."},{"key":"B36","doi-asserted-by":"publisher","first-page":"340","DOI":"10.1111\/j.1542-474x.1999.tb00220.x","article-title":"Noninvasive electrocardiographic imaging","volume":"4","author":"Rudy","year":"1999","journal-title":"Ann. Noninvasive Electrocardiology"},{"key":"B37","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1109\/tassp.1978.1163055","article-title":"Dynamic programming algorithm optimization for spoken word recognition","volume":"26","author":"Sakoe","year":"2003","journal-title":"IEEE Trans. Acoust. Speech, Signal Process."},{"key":"B38","doi-asserted-by":"publisher","first-page":"653013","DOI":"10.3389\/fphys.2021.653013","article-title":"Electrocardiographic imaging for atrial fibrillation: a perspective from computer models and animal experiments to clinical value","volume":"12","author":"Salinet","year":"2021","journal-title":"Front. Physiology"},{"key":"B39","doi-asserted-by":"publisher","first-page":"B605","DOI":"10.1137\/21m1438529","article-title":"Pde-aware deep learning for inverse problems in cardiac electrophysiology","volume":"44","author":"Tenderini","year":"2022","journal-title":"SIAM J. Sci. Comput."},{"key":"B40","doi-asserted-by":"publisher","first-page":"265","DOI":"10.1007\/s10840-015-0068-3","article-title":"Anatomical predictors for successful pulmonary vein isolation using balloon-based technologies in atrial fibrillation","volume":"44","author":"Tsyganov","year":"2015","journal-title":"J. Interventional Cardiac Electrophysiol."},{"key":"B41","doi-asserted-by":"publisher","first-page":"3314","DOI":"10.1093\/eurheartj\/ehae176","article-title":"ESC guidelines for the management of atrial fibrillation developed in collaboration with the european association for cardio-thoracic surgery (EACTS)","volume":"45","author":"Van Gelder","year":"2024","journal-title":"Eur. Heart J."},{"key":"B42","doi-asserted-by":"publisher","first-page":"103093","DOI":"10.1016\/j.artmed.2025.103093","article-title":"Deep learning based estimation of heart surface potentials","volume":"163","author":"Wang","year":"2025","journal-title":"Artif. Intell. Med."},{"key":"B43","doi-asserted-by":"publisher","first-page":"79","DOI":"10.3354\/cr030079","article-title":"Advantages of the mean absolute error (mae) over the root mean square error (rmse) in assessing average model performance","volume":"30","author":"Willmott","year":"2005","journal-title":"Clim. Res."},{"key":"B44","doi-asserted-by":"publisher","first-page":"151","DOI":"10.1109\/tase.2022.3144347","article-title":"Physics-constrained deep learning for robust inverse ecg modeling","volume":"20","author":"Xie","year":"2022","journal-title":"IEEE Trans. Automation Sci. Eng."},{"key":"B45","doi-asserted-by":"publisher","first-page":"129667","DOI":"10.1016\/j.neucom.2025.129667","article-title":"Deep learning for medical imaging super-resolution: a comprehensive review","volume":"630","author":"Xiao","year":"2025","journal-title":"Neurocomputing"},{"key":"B46","doi-asserted-by":"publisher","first-page":"596","DOI":"10.1111\/cpf.12342","article-title":"Atrial average conduction velocity in patients with and without paroxysmal atrial fibrillation","volume":"37","author":"Zheng","year":"2017","journal-title":"Clin. Physiology Funct. Imaging"}],"updated-by":[{"DOI":"10.3389\/fphys.2026.1788430","type":"correction","label":"Correction","source":"publisher","updated":{"date-parts":[[2026,1,26]],"date-time":"2026-01-26T00:00:00Z","timestamp":1769385600000}}],"container-title":["Frontiers in Physiology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fphys.2025.1720244\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,26]],"date-time":"2026-01-26T12:11:18Z","timestamp":1769429478000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fphys.2025.1720244\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,12]]},"references-count":46,"alternative-id":["10.3389\/fphys.2025.1720244"],"URL":"https:\/\/doi.org\/10.3389\/fphys.2025.1720244","relation":{},"ISSN":["1664-042X"],"issn-type":[{"value":"1664-042X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1,12]]},"article-number":"1720244"}}