Multicomponent Organization Analysis in Spatial Domains of Atrial Fibrillation

Atrial fibrillation (AF) is a common heart rhythm disorder associated with elevated health risks. This study uses regression models to identify heart tissue regions linked to AF. We examined the frequency characteristics of atrial areas employing an elastic net regression (ENR) technique to pinpoint significant frequency contributions, creating three-dimensional (3D) maps illustrating the likelihood of arrhythmia origins. We evaluated the effectiveness of our method by applying it to both two-dimensional (2D) and 3D AF simulations and comparing the results with those obtained using least-squares (LS) algorithms. The simulations successfully identified stable rotor and wavebreak regions, though some harmonic frequencies were not captured. We observed defined regional maps in normal atrial tissue, with notable harmonics occurring at 5.9 Hz in the left atrium. The right atrium displayed a smaller rotor region with some missed harmonic frequencies. In fibrotic left atrial substrates, harmonics at 7.8 Hz were consistently detected. Using multi-component domains allowed for a comprehensive analysis, and different estimation methods produced comparable results, facilitating the localization of spatial regions containing AF sources despite a moderate loss of harmonic frequencies.