Non-Invasive Electroanatomical Mapping: A State-Space Approach for Myocardial Current Density Estimation.

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging

Background:

• Cardiac arrhythmias are often treated with invasive catheter ablation.
• Non-invasive methods like electrocardiographic (ECG) and magnetocardiographic (MCG) imaging aim to reconstruct heart's electrical activity.
• Developing non-invasive alternatives to catheter electroanatomical mapping is crucial for reducing patient risk and improving treatment planning.

Purpose of the Study:

• To introduce and mathematically detail a novel system state-based method for estimating human heart electrical activity from MCG measurements.
• To enable arbitrary propagation paths and velocities within the heart's electrophysiological model.
• To lay the groundwork for future research in non-invasive cardiac electrophysiology.

Main Methods:

• Utilized a system state-based approach for electrical activity estimation from MCG data.
• Employed a Kalman filter for optimal estimation of current densities based on measurements and model parameters.
• Implemented gradient descent in an outer optimization loop to refine model parameters.

Main Results:

• Demonstrated the algorithm's feasibility through a simplified one-layer simulation.
• Successfully learned cardiac electrical propagation paths directly from magnetic field measurements.
• Achieved a DICE score of 0.84, recall of 0.77, and precision of 0.93 for tissue segmentation.

Conclusions:

• The developed method shows promise for accurate non-invasive estimation of cardiac electrical activity.
• The algorithm's ability to learn propagation paths and segment tissue highlights its potential for clinical applications.
• This work provides a foundational mathematical framework for advancing non-invasive electroanatomical mapping techniques.