Graph signal processing for multipolar electrogram characterization in catheter ablation of persistent atrial fibrillation
Responsible for a significant proportion of brain strokes, atrial fibrillation (AF) is the most common sustained cardiac arrhythmia encountered in clinical practice. Currently, the most attractive therapeutic option for persistent AF is catheter ablation, where multipolar catheters – i.e., equipped with multiple electrodes – are increasingly used as they facilitate the electroanatomic mapping of the atria before the ablation phase. However, most ablation strategies rely on the prior estimation of the local activation time of the mapped points, neglecting other potentially useful information contained in multipolar atrial electrograms (EGM). In a bid to define a new patient-tailored ablation paradigm, the project aims at analyzing the observed multipolar signals while considering the structural configuration (geometry) of the mapping catheter by means of graph signal processing (GSP). Our focus will be on undirected connected weighted graphs that, unlike their directed counterparts, do not need to infer the causality of the activation process. The underlying hypothesis is that local propagation patterns in AF are associated with specific space-time signatures in the electrograms measured by multipolar catheters, and that such signatures can be captured by suitable GSP tools exploiting the catheter configuration. To test this hypothesis, the project will explore graph characterizations of multipolar EGM to derive biomarkers of local propagation patterns, analyze the ability of the derived biomarkers to identify active drivers sustaining AF, and propose novel patient-tailored ablation protocols relying on the derived graph-based biomarkers. Well-grounded on interpretable mathematical concepts, the GSP approach to be developed in this project is expected to yield reproducible, explainable results, increasing the reliability of clinical decision-making. This interdisciplinary project will be undertaken by an international consortium composed of specialists in signal processing (I3S Lab), computational cardiac modeling (CaMo@KIT) and interventional cardiology (Nice Pasteur CHU, Monaco CHPG).
3/6 complete
Imported from OpenAIRE Germany | Grant ANR-25-CE45-0193.
Open sourceOverview
Grant agreement ANR-25-CE45-0193 | GrAF
Open sourceResponsible for a significant proportion of brain strokes, atrial fibrillation (AF) is the most common sustained cardiac arrhythmia encountered in clinical practice. Currently, the most attractive therapeutic option for persistent AF is catheter ablation, where multipolar catheters – i.e., equipped with multiple electrodes – are increasingly used as they facilitate the electroanatomic mapping of the atria before the ablation phase. However, most ablation strategies rely on the prior estimation of the local activation time of the mapped points, neglecting other potentially useful information contained in multipolar atrial electrograms (EGM). In a bid to define a new patient-tailored ablation paradigm, the project aims at analyzing the observed multipolar signals while considering the structural configuration (geometry) of the mapping catheter by means of graph signal processing (GSP). Our focus will be on undirected connected weighted graphs that, unlike their directed counterparts, do not need to infer the causality of the activation process. The underlying hypothesis is that local propagation patterns in AF are associated with specific space-time signatures in the electrograms measured by multipolar catheters, and that such signatures can be captured by suitable GSP tools exploiting the catheter configuration. To test this hypothesis, the project will explore graph characterizations of multipolar EGM to derive biomarkers of local propagation patterns, analyze the ability of the derived biomarkers to identify active drivers sustaining AF, and propose novel patient-tailored ablation protocols relying on the derived graph-based biomarkers. Well-grounded on interpretable mathematical concepts, the GSP approach to be developed in this project is expected to yield reproducible, explainable results, increasing the reliability of clinical decision-making. This interdisciplinary project will be undertaken by an international consortium composed of specialists in signal processing (I3S Lab), computational cardiac modeling (CaMo@KIT) and interventional cardiology (Nice Pasteur CHU, Monaco CHPG).