A comparison of discretization methods for the inverse problem of electrocardiography

Epicardial electrical activity reconstructed non-invasively from body surface potentials is currently used clinically to map electrical activity. Here, two methods for solving this problem are evaluated using in-vivo experimental data: A coupled Finite-Boundary Element Method (F-BEM) and the Method of Fundamental Solutions (MFS). Results showed similar accuracy, with a slight improvement of the MFS over the F-BEM in reconstructing ventricular electrogram magnitude, (RE
_mean
= 0.83 ± 0.09 and 0.86 ± 0.08 respectively; in the reconstructing activation maps (CC = 0.75±0.09 and 0.73±0.12) and localization of initial activation sites (LE = 19 ± 22 mm and 27±22 mm), though these were still not identified reliably. No improvement was seen with the inclusion of inhomogeneous structures in the F-BEM models (p > 0.05). With the additional advantages in ease of implementation, the MFS is the preferred approach in this case.