The forward and inverse solution during atrial fibrillation: from proof-of-concept to clinical application
Background: Atrial fibrillation (AF) is the sustained cardiac arrhythmia most common in clinical practice, affecting between 1 and 2% of the world population. This disorder has high morbidity and mortality and has become a chronic, non-infectious cardiovascular epidemic that poses a serious threat to human health, making it a crucial public health problem, with a high burden on the National Health Service budget. In the last decades, basic and clinical research has made great progress in improving the diagnosis and treatment of AF, and its mechanism has been gradually elucidated, but not fully understood. During AF, the interpretations of the signals and maps provided by commercial electrical mapping systems is in many cases complex and uncertain, making it difficult to correctly characterize and locate arrhythmogenic sources for ablation. Correct identification of the type of mechanism and its location in the atria is the current challenge for electrophysiologists. Considering the complexity of this arrhythmia and the high sensitivity to errors in current commercial systems, it is important to develop and validate methods for AF diagnostics.
Objective: This study aims to present a customized torso and epicardium electrical mapping for AF patients.
Methods: AF patients underwent magnetic resonance imaging scanning, followed by simultaneous endocardial and body surface electrical mapping. From the body surface signals, the epicardial electrograms were estimated using the non-invasive electrocardiographic imaging (ECGi) method, where torso and atria geometries were obtained from MRI, discretized into triangular elements, followed by the Tikhonov regularization method. From endocardial signals, body surface signals were estimated through the forward solution, considering the torso and atria geometries, and the endocardial electrograms. Torso and atria segmentation were obtained through a pipeline of imaging processing and discretization; then the respective nodes and faces were identified for creation of personalized anatomical geometry from the torso and atria. Time and frequency domain metrics and 3D maps were calculated from the epicardial and body surface electrical signals, allowing for characterization and interpretation of the arrhythmia patterns.
Results: The segmentation of the atria and torso geometries was feasible, allowing for personalized anatomical geometry from the torso and atria from AF patients. In the case of the forward solution, the estimated torso potentials were observed in 3D BSPM AF maps in the time and frequency domain. Analogously, atrial potentials were estimated through the inverse solution, and electrophysiological 3D atrial maps were constructed in the time and frequency domain. The 2D and 3D phase maps were also constructed to visualize the spatiotemporal dynamics of activation and repolarization episodes and to identify the singularity points (SPs), in conjunction with the histogram of the SPs belonging to filaments through heatmaps (HMs). These maps show short filaments in the torso, and a high number of clusters in the SP distribution, which in turn had a small SP density. This strategy was shown to be important for the phase analysis and its biomarkers.
Conclusion: In this study, important and complex concepts about anatomy, cardiac electrophysiology, medical imaging and biomedical signal processing, mathematics, and physics, were unified; this allowed for the proof-of-concept and creation of non-invasive and invasive personalized mapping tools, for studying AF in patients. This proof-of-concept can be extrapolated to other cases of arrhythmias. The refinement and development of new techniques to be implemented as current health technologies represent an innovation for contribution to medical diagnosis and prognosis within medical environments.