The Strategic Research Agenda for cardiovascular disease (SRA-CVD) highlights the improvement of the Atrial Fibrillation (AF) treatment as a priority area. There are more than 6 million AF patients across Europe. Currently, the most effective treatment for AF is ablation. However, more than 60% of persistent AF require several procedures, amounting an excess of €13,5 billion of healthcare costs. Success rates can be doubled if a personalized AF ablation strategy is performed.
Objective. Our goal is to develop a panoramic mapping tool that combines patient-specific electrical and structural information into a single map that accounts for the relevance in the maintenance of AF in each atrial site by taking advantage of current state of the art deep learning strategies.
Methodology. We will: (1) collect a database of clinical data (98 patients) that includes different imaging modalities (i.e. gadolinium enhanced magnetic resonance imaging, non-invasive electrocardiographic imaging and endocardial electroanatomic mapping). From the clinical data, (2) realistic personalized mathematical models will be used to augment the data (2548 subjects) and simulate 12 ablation strategies generating 30.576 ablation scenarios. Thanks to the clinical and mathematical data, we will (3) develop the ESSENCE tool, an interpretable deep-learning methodology to quantify each atrial site according to its predicted contribution to AF maintenance.
Finally, we will (4) validate the developed ESSENCE predicted scores by guiding ablation procedures
in a second group of 98 AF patients.
Team. ESSENCE team is constituted by 15 researchers (4 as full time in the research team and 10 in the working team) from different disciplines: Telecommunication Engineers, Mathematicians, Clinicians and Biomedical Engineers, ensuring the appropriate knowledge, capacities, and experience for the development of the project.
Impact. A technological solution like ESSENCE will increase the efficacy of AF treatments, increasing the number of cured patients and reducing the healthcare costs.
Grant PID2020-119364RB-I00 funded by MCIN/AEI/10.13039/501100011033