Resumen de Estimation of atrial electrical complexity during atrial fibrillation by solving the inverse problem of electrocardiography
Atrial fibrillation (AF) is the most prevalent arrhythmia in the world and is associated with significant morbidity, mortality, and healthcare costs. Despite advancements in pharmaceutical treatment alternatives and ablation therapy, AF management remains suboptimal.
Electrocardiographic Imaging (ECGI) has emerged as a promising non-invasive method for assessing cardiac electrophysiology and guiding therapeutic decisions in atrial fibrillation.
However, ECGI faces challenges in dealing with accurately resolving the ill-posed inverse problem of electrocardiography and optimizing the quality of ECGI reconstructions. Additionally, the integration of ECGI into clinical workflows is still a challenge that is hindered by the associated costs arising from the need for cardiac imaging.
For this purpose, the main objectives of this PhD thesis are to advance ECGI technology by determining the minimal technical requirements and refining existing methodologies for acquiring accurate ECGI signals. In addition, we aim to assess the capacity of ECGI for noninvasively quantifying AF complexity. To fulfill these objectives, several studies were developed throughout the thesis, advancing from ECGI enhancement to AF evaluation using ECGI.
Firstly, geometric and signal requirements of the inverse problem were addressed by studying the effects of torso mesh density and electrode distribution on ECGI accuracy, leading to the identification of the minimal number of nodes and their distribution on the torso mesh. Besides, we identified that the correct location of the electrodes on the reconstructed torso mesh is critical for the accurate ECGI signal obtention. Additionally, a new methodology of imageless ECGI was defined and assessed by comparing ECGI-derived drivers computed with the original heart geometry of the patients to the drivers measured in different heart geometries. Our results showed the ability of imageless ECGI to the correct quantification and location of atrial fibrillation drivers, validating the use of ECGI without the need for cardiac imaging. Also, the current state of-the-art zero-order Tikhonov regularization and L-curve optimization for computing ECGI signals were improved by investigating the impact of electrical noise and geometrical uncertainties on the regularization. We proposed a new criterion that enhances the accuracy and reliability of ECGI solutions in situations with uncertainty from unfavorable signal conditions.
Secondly, in this PhD thesis, several analyses, signal processing methodologies, and ECGIderived metrics were investigated to better characterize the cardiac substrate and reentrant activity in ECGI signals from AF patients. With the objective of obtaining a deeper understanding of the electrophysiological mechanisms underlying AF, we established the optimal filtering strategy to extract patient-specific reentrant patterns and derived metrics in ECGI signals. Furthermore, we investigated the reproducibility of the obtained ECGI-reentrant maps and linked them to the success of PVI ablation. Our results showed that higher reproducibility on AF drivers detected with ECGI is linked with the success of PVI, creating a proof-of-concept mechanism for stratifying AF patients prior to ablation procedures.
