Abstract
Patient-specific myocardial conduction velocity (CV) is important for cardiac electrophysiology modelling but is commonly assigned using population-level values. We investigated whether compact QRS slope dynamics from surface ECG could predict electroanatomical mapping (EAM)-derived CV estimates for simulation parameterisation. In 23 clinical cases, sinus-rhythm QRS waveforms from leads V3-V5 were converted into compact dV/dt-based representations and combined with timing and morphology descriptors to generate 165 case-level features. A regularised masked multi-task multilayer perceptron predicted epicardial longitudinal, endocardial longitudinal, and exploratory apparent transmural CV while handling missing labels through a target-masked loss. Performance was assessed using leave-one-case-out cross-validation against a training-set mean baseline and target-specific XGBoost regressors. The neural model reduced mean absolute error compared with the baseline for epicardial longitudinal CV (16.83 vs 20.32 cm/s), endocardial longitudinal CV (10.05 vs 15.73 cm/s), and apparent transmural CV (3.01 vs 5.15 cm/s), with corresponding R2 values of 0.072, 0.417, and 0.406. A nested sensitivity analysis of network architecture and regularisation produced similar MAEs of 16.62, 9.76, and 3.17 cm/s. These internal results support feasibility but require cautious interpretation and independent validation.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/STACOM2026_021.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/profile?id=~Fuyu_Cheng1
BibTex
@InProceedings{CheFuy_ECGBased_MICCAISAT2026,
author = { Cheng, Fuyu AND Tonko, Johanna B. AND Rauseo, Elisa AND Misghina, Semhar B. AND Barrera, Carlos E. AND Chen, Xu AND Vigmond, Edward J. AND Slabaugh, Gregory AND Aung, Nay AND Petersen, Steffen E. AND Lambiase, Pier D. AND Roney, Caroline H.},
title = { { ECG-Based Prediction of Patient-Specific Myocardial Conduction Velocity with Electroanatomical Mapping Supervision } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17267},
month = {pending},
page = {pending}
}