Identifying the minimum electrode count for locating potential atrial fibrillation drivers in electrocardiographic imaging using unsupervised electrode placement

Objective: To promote the effectiveness in measuring high-density body surface potentials (BSPs) and optimize the clinical application of electrocardiographic imaging (ECGI) to identify atrial fibrillation (AF) drivers, it is essential to identify a minimal electrode number.

Methods: This study included 24 participants with paroxysmal or persistent AF. We compared the reconstructed entropy maps and epicardial potential maps using 10 different electrode counts, ranging from 12 to 128, to determine the minimum necessary number of electrodes. A novel atrial activity extraction method and an improved spectral clustering technique were designed to intelligently identify electrode locations from the 128 evenly distributed electrodes.

Results: For patients with persistent AF, the correlation coefficient (CC) of the reconstructed potential maps ranged from 0.72 to 0.99, and the structural similarity index (SSIM) >0.90 when the number of electrodes was ≥64. For patients with paroxysmal AF, the CC ranged from 0.85 to 0.99 and the SSIM≥0.90 when the number of electrodes was ≥48. Across all 24 AF patients, when electrode numbers were ≥64, the mean ± standard deviation of the CC between pairwise potential maps was >0.93 ± 0.06. Electrode placement on the upper anterior and posterior regions of the torso, above the sixth intercostal septum, was found to be advantageous for capturing atrial electrical activity conducive to identifying AF drivers.

Conclusion: For patients with persistent AF, ≥64 electrodes are recommended. For patients with paroxysmal AF, ≥48 electrodes are recommended.