Evaluation of Computed Epicardial Potentials, Activation Times, and Pacing Site Localization in Electrocardiographic Imaging

Electrocardiographic imaging (ECGI) addresses the challenge of reconstructing cardiac electrical sources from body-surface potential distributions, known as the inverse problem of electrocardiography. The Consortium for ECG Imaging (CEI) attempts to transition ECGI from research to clinical practice by investigating the impacts of features such as signal preprocessing and forward model accuracy on inverse reconstruction. Using data from a human-shaped torso tank and a canine heart model, this study involves a multi-step computational approach. MRI images provide anatomical context, while ECGs capture electrical activities on the body's surface. Zero-order Tikhonov regularization stabilizes the inverse problem solution, enabling accurate epicardial potential computation. Activation times are estimated through a spatiotemporal method that integrates both spatial and temporal changes in cardiac potentials, and pacing sites are localized based on these activation times. Evaluation of the reconstructed electrograms (EGMs), activation times (ATs), and pacing site localization showed strong agreement with measured data. Median correlation coefficients for EGMs ranged from 0.74 to 0.82, and for ATs from 0.84 to 0.90, indicating strong agreement. Localization errors varied, with the smallest error at the LV-apex (5.64 mm) and the largest at the RV (15.17 mm). These results underscore the effectiveness of the ECGI framework and highlight areas for further refinement to enhance clinical applicability.