Automated detection and characterization of complex fractionated atrial electrograms in human left atrium during atrial fibrillation

Background: Complex fractionated atrial electrograms (CFAEs) have been reported as ablative targets for the treatment of atrial fibrillation (AF). However, the process of CFAE identification is highly dependent on the operator's judgment. Objective: It is the aim of the study to report our initial experience with a novel software algorithm designed to automatically detect CFAEs. Methods: Nineteen patients (6 female, 58 ± 8 years) who underwent catheter ablation of paroxysmal (n = 11) or persistent (n = 8) AF were included in the study. During ongoing AF, 100 ± 15 left atrial (LA) endocardial locations were sampled under the guidance of integrated electroanatomical mapping with computed tomographic images. Bipolar electrograms recorded throughout the LA were analyzed using custom software that allows for automated detection of CFAEs. Interval confidence level (ICL), defined as the number of intervals between consecutive CFAE complexes during 2.5-second recordings, was used to characterize CFAEs. The CFAE sites with an ICL ≥5 were considered as sites with highly repetitive CFAEs, which are thought to be potential ablation targets. For purposes of analysis, the LA was divided into 6 areas: pulmonary vein (PV) ostia, posterior wall, interatrial septum, roof, mitral annulus area, and appendage. Results: Among a total of 1,904 LA locations sampled in 19 patients, 1,644 (86%) were categorized as CFAE sites, whereas 260 (14%) were categorized as as non-CFAE sites. Thirty-four percent of all CFAE sites were identified as sites with highly repetitive CFAEs. Of these, 24% were located at the interatrial septum, 22% on the posterior wall, 20% at the PV ostia, 18% at the mitral annulus area, 14% on the roof, and 2.7% at the LA appendage. In all patients, highly repetitive CFAE sites were distributed in 4 or more areas of the LA. Persistent AF patients had more highly repetitive CFAE sites on the posterior wall than paroxysmal AF patients (30% ± 7.3% vs 14% ± 8.2%, P < .001). There was a strong trend toward more highly repetitive CFAE sites located at the PV ostia in patients with paroxysmal AF compared with persistent AF patients (24% ± 13% vs 13% ± 7.7%, P = .05). Conclusion: With the use of custom software, CFAE complexes were identified in more than 80% of the LA endocardial locations. LA sites with highly repetitive CFAE sites were located predominately in the septum, posterior wall, and PV ostia. Patients with persistent AF had a different anatomical distribution pattern of highly repetitive CFAE sites from those with paroxysmal AF, with a greater prevalence of highly repetitive CFAEs located on the posterior wall. Further studies are warranted to determine the clinical significance of these findings.