Automatic Extraction of Recurrent Patterns of High Dominant Frequency Mapping During Human Persistent Atrial Fibrillation

Author(s):Show less -

Li, Xin ^{[1, 2]} ; Chu, Gavin S. ^[1] ; Almeida, Tiago P. ^{[1, 2]} ; Vanheusden, Frederique J. ^[3] ; Salinet, Joao ^[4] ; Dastagir, Nawshin ^[5] ; Mistry, Amar R. ^[1] ; Vali, Zakariyya ^{[1, 6]} ; Sidhu, Bharat ^[1] ; Stafford, Peter J. ^[6] ; Schlindwein, Fernando S. ^{[6, 2]} ; Ng, G. Andre ^{[1, 6]}

Affiliation

^[1] Univ Leicester, Dept Cardiovasc Sci, Leicester, Leics - England

^[2] Univ Leicester, Sch Engn, Leicester, Leics - England

^[3] Nottingham Trent Univ, Sch Sci & Technol, Nottingham - England

^[4] Fed Univ ABC, Ctr Engn Modelling & Appl Social Sci CECS, Biomed Engn, Santo Andre, SP - Brazil

^[5] Univ Auckland, Auckland Bioengn Inst, Auckland - New Zealand

^[6] Glenfield Hosp, Natl Inst Hlth Res, Leicester Cardiovasc Biomed Res Ctr, Leicester, Leics - England

• Document type: Journal article
• Source: FRONTIERS IN PHYSIOLOGY; v. 12, MAR 12 2021.
• Web of Science Citations: 0

Abstract

Purpose: Identifying targets for catheter ablation remains challenging in persistent atrial fibrillation (persAF). The dominant frequency (DF) of atrial electrograms during atrial fibrillation (AF) is believed to primarily reflect local activation. Highest DF (HDF) might be responsible for the initiation and perpetuation of persAF. However, the spatiotemporal behavior of DF remains not fully understood. Some DFs during persAF were shown to lack spatiotemporal stability, while others exhibit recurrent behavior. We sought to develop a tool to automatically detect recurrent DF patterns in persAF patients. Methods: Non-contact mapping of the left atrium (LA) was performed in 10 patients undergoing persAF HDF ablation. 2,048 virtual electrograms (vEGMs, EnSite Array, Abbott Laboratories, USA) were collected for up to 5 min before and after ablation. Frequency spectrum was estimated using fast Fourier transform and DF was identified as the peak between 4 and 10 Hz and organization index (OI) was calculated. The HDF maps were identified per 4-s window and an automated pattern recognition algorithm was used to find recurring HDF spatial patterns. Dominant patterns (DPs) were defined as the HDF pattern with the highest recurrence. Results: DPs were found in all patients. Patients in atrial flutter after ablation had a single DP over the recorded time period. The time interval (median {[}IQR]) of DP recurrence for the patients in AF after ablation (7 patients) decreased from 21.1 s {[}11.8 49.7 s] to 15.7 s {[}6.5 18.2 s]. The DF inside the DPs presented lower temporal standard deviation (0.18 +/- 0.06 Hz vs. 0.29 +/- 0.12 Hz, p < 0.05) and higher OI (0.35 +/- 0.03 vs. 0.31 +/- 0.04, p < 0.05). The atrial regions with the highest proportion of HDF region were the septum and the left upper pulmonary vein. Conclusion: Multiple recurrent spatiotemporal HDF patterns exist during persAF. The proposed method can identify and quantify the spatiotemporal repetition of the HDFs, where the high recurrences of DP may suggest a more organized rhythm. DPs presented a more consistent DF and higher organization compared with non-DPs, suggesting that DF with higher OI might be more likely to recur. Recurring patterns offer a more comprehensive dynamic insight of persAF behavior, and ablation targeting such regions may be beneficial. (AU)

• FAPESP's process: 17/00319-8 - Atrial substrate identification in patients with chronic atrial fibrillation using multivariate statistical models and multiple attributes from atrial electrograms
• Grantee: Tiago Paggi de Almeida
• Support Opportunities: Scholarships in Brazil - Post-Doctoral