Bayesian Framework for Atrial LAT Estimation in ECGI.

Area of Science:

• Cardiology
• Biomedical Engineering
• Computational Biology

Background:

• Local activation time (LAT) mapping provides crucial insights into cardiac electrical propagation.
• Traditional LAT estimation methods using invasive systems create artifacts when applied to non-invasive ECGI signals.
• Existing non-invasive methods like -dV/dt and spatiotemporal gradient (STG) have limitations in accuracy and artifact generation.

Purpose of the Study:

• To introduce and evaluate a novel Bayesian framework for estimating LATs from ECGI signals.
• To compare the performance of the Bayesian framework against traditional -dV/dt and STG methods.
• To assess the framework's ability to accurately map cardiac electrical activity and identify lines of block (LoB) without artifacts.

Main Methods:

• Development of a novel Bayesian framework for LAT estimation using ECGI data.
• Validation using in-silico pacing models with varying noise levels.
• Clinical evaluation on two patients: one in sinus rhythm and one post-cavotricuspid isthmus (CTI) ablation during pacing.

Main Results:

• The Bayesian approach achieved Pearson correlation coefficients >0.91 in simulations, significantly outperforming -dV/dt (0.75-0.81) and STG (0.78-0.86) across noise levels.
• Reduced error in identifying earliest and latest activation sites by up to 0.8 cm in low-noise conditions.
• Clinically, the Bayesian framework accurately identified real LoB and reduced artificial LoB by an average of 1% in simulation data.

Conclusions:

• The novel Bayesian framework offers a significant advancement for non-invasive LAT estimation using ECGI.
• This method provides accurate, artifact-free cardiac mapping, outperforming conventional techniques.
• It represents a promising alternative for the efficient and precise mapping of cardiac arrhythmias.