Hierarchical Algorithms for Causality Retrieval in Atrial Fibrillation Intracavitary Electrograms

Area of Science:

• Cardiology
• Biomedical Engineering
• Computational Neuroscience

Background:

• Atrial fibrillation (AF) treatment relies on radiofrequency catheter ablation guided by multichannel intracavitary electrograms (EGMs).
• Identifying critical ablation targets from complex EGM data remains a challenge for cardiologists.

Purpose of the Study:

• To develop novel hierarchical algorithms for inferring causal interactions among multichannel EGMs.
• To enhance the understanding of electrical signal propagation patterns in the atria for improved AF ablation guidance.

Main Methods:

• Introduction of two Granger causality-based hierarchical algorithms: GS-CaRe (global search) and LS-CaRe (local search).
• Algorithms analyze multichannel intracavitary electrograms to establish cause-effect relationships between signal sources.
• Validation using both synthetic and real-world annotated electrophysiological signals.

Main Results:

• Both GS-CaRe and LS-CaRe algorithms successfully retrieved causal interactions from multichannel EGMs.
• The causal models provided insights into wavefronts and activation patterns within the atria.
• Demonstrated good performance in numerical experiments on diverse datasets.

Conclusions:

• The proposed algorithms offer a robust method for analyzing complex EGM data.
• Causal modeling of EGMs can significantly aid cardiologists in pinpointing optimal targets for radiofrequency catheter ablation in AF patients.