Abstract
The goal of this study was to identify the spatial resolution requirements for accurate rotor detection and localization in human right ventricular tachyarrhythmias. Poor spatial resolution is often cited as a reason for the inaccuracy of cardiac mapping catheters in detecting and localizing arrhythmia rotors. High-resolution (0.7 mm) arrhythmia data from optical recordings obtained from human donor hearts (n = 12) were uniformly downsampled to lower resolutions (1.4-7 mm) to approximate the spatial resolution (4 mm) of clinical mapping catheters. Rotors were tracked at various subresolutions and compared to the rotors in the original data by computing F1-scores to create accuracy profiles for both rotor detection and localization. Further comparisons were made according to arrhythmia type, donor sex, anatomical region, and mapped surface: endocardium or epicardium. For a spatial resolution of 4.2 mm, the accuracies of rotor detection and localization were 57% ± 4% and 61% ± 7%, respectively. Arrhythmia type affected the accuracy of rotor detection (monomorphic ventricular tachycardia, 58% ± 4%; ventricular fibrillation, 56% ± 8%) and localization (monomorphic ventricular tachycardia, 70% ± 4%; ventricular fibrillation, 54% ± 13%). However, donor sex, anatomical region (right ventricular outflow tract, mid, and apical), and mapped surface (epicardium and endocardium) did not significantly affect rotor detection or localization accuracy. To achieve rotor detection accuracy of 80%, a spatial resolution of 1.4 mm or better is needed. The accuracy profiles provided here serve as a guideline for future mapping device development.