Performance of Iterative Methods of ECGI in the Atria

The data completion problem of electrocardiographic
imaging (ECGI), i.e., computing pericardial potentials
from the body surface potentials, is a challenging task. In
this context, many different methodologies can be used to
solve this problem. These are divided into direct method
(Tikhonov regularization) and indirect methods (iterative
methods).
Employing realistic simulations, we compare three iterative methods: the Landweber method, the Algebraic
Reconstruction Technique (ART), and the Range Restricted Generalized Minimal Residual method (RRGMRES), against the standard Zero Order Tikhonov regularization. Our analysis focuses on five atrial ectopics and
five atrial reentries simulations, assessing the methodologies based on Correlation Coefficient (CC) of electrograms
(EGMs) and Local Activation Times (LATs).
Iterative methods show promise in ECGI, obtaining
equal or better results than Tikhonov regularization. However, due to the need of an independent stopping criteria
Tikhonov regularization still leads on real applicability,
suggesting its continued value in cardiac diagnostic applications despite the emerging potential of iterative approaches.