Electrocardiographic imaging and phase mapping approach for atrial fibrillation: A simulation study

Objectives: We aimed to evaluate the non-invasive phase mapping methods used in clinical practice on atrial signals during atrial fibrillation (AF). Methods: A modified Courtemanche human atrial ionic model was used to run AF simulations. Extracellular potentials on the epicardium were computed and propagated to the body surface through a homogeneous torso conductor using the Boundary Element Method. The obtained body surface potentials were sampled in 252 different locations to replicate clinical recordings. The clinical non-invasive AF mapping workflow was then applied to this body surface data to reconstruct atrial epicardial potentials and corresponding phase signals. Results: The AF cycle lengths were well estimated for the two datasets (mean relative error magnitude MRE=5.4% and 3.8% for the two simulation sequences with no noise). Results were maintained when up to 10 dB of signal noise on the body surface recordings or 7.5+/−3.4mm geometrical noise on the electrode locations were added. The phase locking values (PLV) were 0.62 and 0.78 respectively for the two simulation sequences, indicating a fair correlation between the phase signals. Regions showing reentries were correctly localized. Reconstructed phase singularity positions were insensitive to added electrical and geometrical noise.