Introduction: Despite recent advancements, Electrocardiographic imaging (ECGI) systems are still plagued by a myriad of errors, which makes studying and improving these systems difficult. Many of these errors arise from the signal and geometry acquisition, including registration errors, noisy signals, and tissue inhomogeneities. To mitigate these errors, we developed a novel rigid pericardiac recording array for a torso tank experimental preparation and tested several ECGI approaches.
Methods: A 256-electrode rigid pericardiac recording array was designed to record signals 0.5--1.0~cm above the entire surface of an isolated heart. The cage and heart were lowered into a 192-electrode torso tank array. The heart was paced from the anterior and posterior left ventricular free wall, and 40 beats from each site were selected. We used this preparation to examine several standard ECGI techniques: boundary element method (BEM) with the Tikhonov zero-order (Tik0), Tikhonov second-order (Tik2), and truncated singular value decomposition (TSVD) regularization and the method of fundemental solutions (MFS) with Tik0 regularization. Regularization parameters were selected using the L-curve method for BEM and CRESO method for MFS.
Results: Each ECGI formulation performed well, as observed by spatial correlation (above 0.7), temporal correlation (above 0.8), and root mean squared error (below 0.7). Of the four inverse methods, MFS performed the best, with mean spatial and temporal correlations of 0.90 and 0.95, respectively. The localization of the earliest site of activation resulted in higher error than expected for Tik0 and TSVD, with an mean range of 2.0-3.6 cm.
Discussion: Our novel experimental preparation, using the pericardiac cage, significantly limited the error resulting from geometry, conductivity, and signal acquisition. To our knowledge, the ECGI performance was the highest published with an experimental validation model and have shown the utility to compare inverse techniques. Future work will use this experimental preparation to validate and improve ECGI systems.