Data-driven estimation of cardiac electrical diffusivity from 12-lead ECG signals

Area of Science:

• Computational biology
• Medical physics
• Cardiology

Background:

• Dilated cardiomyopathy (DCM) diagnosis and treatment are complex due to diverse causes and stages.
• Personalized computational models of cardiac electrophysiology (EP) are crucial for improving DCM assessment, prognosis, and therapy planning.
• Current EP models require patient-specific parameters, often difficult to obtain.

Purpose of the Study:

• To develop a data-driven approach for estimating electrical diffusivity parameters in cardiac EP models.
• To personalize EP models using standard 12-lead electrocardiograms (ECGs).
• To assess the feasibility and accuracy of this personalization method for DCM patients.

Main Methods:

• Utilized an efficient mono-domain Lattice-Boltzmann model for cardiac EP simulations.
• Employed a boundary element method for mapping body surface potentials.
• Estimated electrical diffusivity using polynomial regression based on QRS duration and electrical axis from ECGs.

Main Results:

• Successfully computed 9500 EP simulations for 19 DCM patients, learning the regression model efficiently.
• Quantified the uncertainty in electrical diffusion prediction based on ECG features.
• Achieved 84% personalization success rate in DCM cases, with prediction errors within clinical acceptability (18.7ms for QRS duration, 6.5° for electrical axis).

Conclusions:

• The data-driven approach provides a feasible method for estimating crucial diffusion parameters from readily available clinical ECG data.
• This technique serves as a foundational step towards comprehensive personalization of cardiac EP models.
• Improved personalization can lead to better DCM patient assessment, prognosis, and tailored therapeutic strategies.