Accuracy of inverse solution computation of dominant frequencies and phases during atrial fibrillation

Ablation of sources identified by dominant frequency (DF) or phase analyses in patients with atrial fibrillation (AF) is proposed as a therapy to terminate the arrhythmia. The aim of this study was to evaluate noninvasive identification of AF sources by solving the inverse problem. Realistic mathematical models of atria and torso anatomy with different DF patterns of AF were used. For these models inverse-computed electrograms EGMs were compared to intracardiac EGMs in terms of their voltage, phase and frequency spectrum errors. In models without added noise, atrial spectrums were estimated with lower relative errors than phases (3.8±3.8 vs 41±17 % respectively, p<;0.01) and after noise addition differences between relative errors of spectrums and phases were more pronounced (5.5±4.1 vs 48±14 %, respectively p<;0.01). Non-invasive location of the highest DF during AF showed a better accuracy than in the phase and voltage domain, being a promising clinical tool for identification of sources prior to ablation.