Reconstructing synthetic hearts from ECG using flow matching
- 1Imperial College London
Abstract
Cardiac imaging enables quantitative assessment of cardiac structure and function but remains constrained by cost, infrastructure and specialist expertise. In contrast, electrocardiogram (ECG) is widely accessible yet underexploited, despite encoding latent information about cardiac physiology. Here we introduce visionECG, a conditional flow matching framework that learns a probabilistic mapping between two biological distributions--the space of cardiac electrical signals and the space of cardiac geometries. Using 71,132 paired ECG and cardiac mesh sequence datasets from the UK Biobank, with external assessment in 5,000 patients with ECG-echocardiogram pairs, the model reconstructs quantitatively accurate spatiotemporal representations of the left ventricle using ECG inputs and basic demographic information alone. These reconstructions enable discrimination of structural abnormalities and disease labels, provide visualisations of functional abnormalities, and support flexible quantification of both global and regional parameters. By reframing the ECG as a generative source of patient-specific left ventricular geometry and motion, this work establishes a scalable framework for translating low-dimensional signals into high-dimensional, physiologically grounded structured representations.
Related
- LDCT-to-SDCT as a Bridge Problem: Single-Step Residual Endpoint Flow Matching for Real-Time Denoising
- GliomaINK: A Tumor and Anatomy Conditioned Flow Matching Framework for 3D MRI Sequence Synthesis in Glioma Patients
- Explained and Residual Variability of the Fetal Mechanical PR Interval in Structurally Normal Hearts: A Secondary Analysis of Physiological, Biometric, and Doppler Covariates
- Polyurethane versus pericardium: Contrasting hemocompatibility in 2 different total artificial hearts.
- Head-to-head comparison of central and peripheral venoarterial ECMO in normal ovine hearts and in cardiogenic shock
BibTeX
@misc{zheng2026reconstructing,
title = {{Reconstructing synthetic hearts from ECG using flow matching}},
author = {J. Zheng and S. Kalaie and Q. Ma and Q. Meng and K. Rjoob and P. Gifani and L. Hu and N. Babazade and M. Coriano and W. Zhong and M. Vafaeezadeh and S. Tahasildar and N. Vadgama and D. S. Senevirathne and A. Santhirasekaram and K. A. McGurk and L. Curran and Y. He and L. Chen and Y. Mo and L. Huang and M. Qiao and Y. Huang and W. Bai and D. P. O'Regan},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.01.26360987},
url = {https://www.medrxiv.org/content/10.64898/2026.09.01.26360987v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Reconstructing synthetic hearts from ECG using flow matching
AU - J. Zheng
AU - S. Kalaie
AU - Q. Ma
AU - Q. Meng
AU - K. Rjoob
AU - P. Gifani
AU - L. Hu
AU - N. Babazade
AU - M. Coriano
AU - W. Zhong
AU - M. Vafaeezadeh
AU - S. Tahasildar
AU - N. Vadgama
AU - D. S. Senevirathne
AU - A. Santhirasekaram
AU - K. A. McGurk
AU - L. Curran
AU - Y. He
AU - L. Chen
AU - Y. Mo
AU - L. Huang
AU - M. Qiao
AU - Y. Huang
AU - W. Bai
AU - D. P. O'Regan
PY - 2026
DA - 2026/09/01
PB - medRxiv
DO - 10.64898/2026.09.01.26360987
UR - https://www.medrxiv.org/content/10.64898/2026.09.01.26360987v1
AB - Cardiac imaging enables quantitative assessment of cardiac structure and function but remains constrained by cost, infrastructure and specialist expertise. In contrast, electrocardiogram (ECG) is widely accessible yet underexploited, despite encoding latent information about cardiac physiology. Here we introduce visionECG, a conditional flow matching framework that learns a probabilistic mapping between two biological distributions--the space of cardiac electrical signals and the space of cardiac geometries. Using 71,132 paired ECG and cardiac mesh sequence datasets from the UK Biobank, with external assessment in 5,000 patients with ECG-echocardiogram pairs, the model reconstructs quantitatively accurate spatiotemporal representations of the left ventricle using ECG inputs and basic demographic information alone. These reconstructions enable discrimination of structural abnormalities and disease labels, provide visualisations of functional abnormalities, and support flexible quantification of both global and regional parameters. By reframing the ECG as a generative source of patient-specific left ventricular geometry and motion, this work establishes a scalable framework for translating low-dimensional signals into high-dimensional, physiologically grounded structured representations.
ER -