Generation of cardiac digital twins based on noninvasive cardiac mapping

Abstract

Abstract Funding Acknowledgements Type of funding sources: Other. Main funding source(s): Generalitat Valenciana Grants Ministerio de Ciencia, Innovación y Universidades Background Personalized cardiac digital twins have been proposed as a clinical tool to plan ablation strategies during cardiac arrhythmias. However, these personalization frameworks normally rely on invasive recordings which limits their clinical applicability. Electrocardiographic Imaging (ECGI) is a mapping technique that allows the reconstruction of global cardiac activity non-invasively. Purpose Evaluate the applicability of ECGI as a non-invasive tool for the planification of ablation strategies based on personalized digital twins. Methods Cardiac digital twins generated from endocardial or ECGI mapping were compared in a patient with Atrial Flutter prescribed for cardiac ablation. Endocardial mapping was obtained in the right atrium by using electroanatomical mappings. ECGI was recorded during the ablation procedure with a mapping system of 128 body surface recordings. For the cardiac digital twin based on endocardial mapping, the reentrant circuit around the tricuspid valve was reproduced based on the activation time maps obtained with the electroanatomical mapping system. Total activation time of both atria was personalized based on the duration of the F-waves on the ECG since the left atrium was not endocardialy mapped. For the cardiac digital twin based on the ECGI, activation times obtained in both left and right atrium were used to set-up the personalized model. From both approaches, computed torso potentials were calculated. The correlation with real body surface recordings was compared. To further evaluate the fidelity of the models, the efficacy of the ablation strategy performed on the real patient was evaluated over the digital twins. Results Both digital twins, the one based on endocardial mapping, and the one based on ECGI were compared. Regarding dynamics of the model, both models reproduced the reentrant circuit around the tricuspid valve. Interestingly, only the ECGI digital twin reproduced a reentrant wave around the right pulmonary veins. When endocardial map was used for the personalization, a median conduction velocity (CV) of 90.5 cm/s was obtained for the right atrium and 174.7 cm/s for the left; whereas when ECGI was used, the median CV were 104.2 cm/s at the right and 129.1 cm/s at the left atrium. The ability to reproduce body surface ECGs of both digital twins demonstrated that the one based on endocardial mapping obtained a low correlation of 0.13, whereas the ECGI digital twin increase the correlation up to 0.66. Regarding the ablation outcome, ablation terminated the arrhythmia only in the cardiac digital twin based on ECGI. Conclusion The possibility to generate cardiac digital twins based on non-invasive ECGI mapping has been demonstrated. This opens the door to obtaining personalized cardiac digital twins before endocardial interventions which can be used to plan ablation procedures.