Abstract
Cardiacelectrophysiologicalmodelingandinversereconstruction are fundamental to cardiac digital twins and precision medicine. However, traditional finite element simulations are computationally expensive, and existing black-box surrogates often lack physically interpretable intermediate representations in ill-posed ECG inverse problems. Accordingly,weproposeaninterpretablemulti-taskphysiologicallyadaptive surrogate network that integrates a physiologically adaptive state space block, energy-constrained attention, and physics-conditioned constraints for efficient mappings from physiological parameters to transmembrane potentials and ECGs. A two-stage optimization strategy is further introduced for stable inverse inference of physiological parameters. Experiments show that our model achievesR2 = 0.99in forward inference with approximately 2 ms inference time. In clinical 8-lead ECG inversion, the proposed method shows strong agreement with measured data (Pearson correlation coefficient = 0.93) and reduces optimization time from tens of hours to minutes. These results demonstrate an effective balance among efficiency, expressiveness, and interpretability for rapid cardiac digital twins.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/DT4H_029.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=3TZPYVebS8
BibTex
@InProceedings{LyuYil_Rapid_MICCAISAT2026,
author = { Lyu, Yilin AND Wang, Mengxiao AND Cang, Dongcheng AND Vu, Tran Anh AND Camps, Julia AND Sia, Ching-Hui AND Chan, Mark Y. AND Li, Lei},
title = { { Rapid Electrophysiology Simulation and Inverse Optimization based on Multi-Task Surrogate Modeling for Cardiac Digital Twins } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17275},
month = {pending},
page = {pending}
}