Region specific modeling of cardiac muscle: comparison of simulated and experimental potentials.

This article investigates the quantitative predictive capabilities of region-specific models by comparing experimental electrograms obtained from in vivo mapping of the ventricular free wall with those obtained through simulation of a region specific three-dimensional bidomain model that incorporates measured fiber orientations. Epicardial electrograms were recorded from canine left ventricles during and after unipolar pacing using a 528-channel electrode plaque. Fiber directions throughout the tissue were estimated from diffusion-weighted MRI and from pace mapping. Electrograms were computed in the bidomain model with experimentally derived properties during paced activations at the same spatiotemporal resolution as those recorded experimentally. Epicardial potentials from model and experiment were directly compared, and sensitivities of these comparisons to reference electrode location and to the choice of material properties were analyzed. The comparisons performed here demonstrate, that (1) the stimulus artifact can be used to estimate the in vivo myocardial fiber architecture, (2) the correlation between simulated and experimental electrograms decreases with increasing pacing depth, and (3) the quantitative comparisons between bidomain model and experimental data are sensitive to both the description of the fiber architecture, and the location of the unipolar reference electrode, but relatively insensitive to moderate changes in the bidomain conductivities.