ABSTRACT
Atrial fibrillation (AF) is often successfully treated by catheter ablation. Those cases of AF that do not readily succumb to ablation therapy would benefit from improved methods for mapping the complex spatial patterns of tissue activation that typify recalcitrant AF. To this end, the purpose of our study was to investigate the use of numerical deconvolution to improve the spatial resolution of activation maps provided by 2-D arrays of intra-cardiac recording electrodes. We simulated tissue activation patterns and their corresponding electric potential maps using a computational model of cardiac electrophysiology, and sampled the maps over a grid of locations to generate a mapping data set. Following cubic spline interpolation, followed by edge-extension and windowing, we deconvolved the data and compared the results to the model current density fields. We performed a similar analysis on voltage-sensitive dye maps obtained in isolated sheep hearts. For both the synthetic data and the voltage-sensitive dye maps, we found that deconvolution led to visually improved map resolution for arrays of 10×10 up to 30×30 electrodes placed within a few mm of the atrial surface when the activation patterns included 3-4 features that spanned the recording area. Root mean square error was also reduced by deconvolution. Deconvolution of arrays of intracardiac potentials, preceded by appropriate interpolation and edge processing, leads to potentially useful improvements in map resolution that may allow more effective assessment of the spatiotemporal dynamics of tissue excitation during AF.
