Analysis of the Time-Dependent Behaviors of Atrial Fibrillation with Electrographic Flow Mapping

Abstract

Background Electrographic flow (EGF) mapping algorithms employing Horn-Schunck flow estimations can create temporospatial visualizations of atrial electrical wavefront propagations during atrial fibrillation (AF). Reproducible patterns of centrifugal EGF activation from discrete sites may represent sites of AF origin or sources. Our objectives were to assess the patterns and prevalence of AF sources using EGF mapping. Methods Unipolar electrograms were recorded for 1-minute with 64-pole basket catheters. Flow estimates were constructed by passing consecutive frames through an algorithm to learn and then compare typical wave direction patterns to describe flow-field evolution. During each 2-second segment, sites initiating centrifugal activation patterns were defined as AF sources. Maps of source location/activity duration were generated. Results The EGF method was applied to 405 prospective and retrospective patients with persistent or long-standing persistent AF. Mean age 62.5 years; mean LA size 54 mm; mean AF duration 4.6 years. EGF mapping found 6.6 ± 2.4 AF sources/patient (range 1 to 17). Distribution was 55% LA and 45% RA. Dominant sources (prevalence >20%) were demonstrated in 185 (45.7%) patients, but only 10.7% of all sources were dominant. While AF cycle length (CL) was not affected by source prevalence, CL variance significantly decreased as source prevalence increased. Conclusions Complex AF conduction patterns make ablation challenging, but EGF mapping enables detection and organization of time-dependent AF behaviors. Although many low prevalence sources are detected, they may not be clinically relevant, while higher prevalence sources seem to modulate AF. Recording durations of 1 minute facilitate source discrimination.