Accuracy of estimates of cardiac action potential duration from extracellular waveforms simulated by the Bidomain model

The goal of the present work is to provide an extensive quantitative analysis of the accuracy of activation - recovery intervals (ARIs) derived from unipolar electrograms (EGs) and hybrid monophasic action potentials (HMAPs) under normal and ischemic conditions. ARIs are compared with the gold standard action potential durations (APDs) based on the transmembrane action potential (TAP). This study is based on large scale parallel 3D numerical simulations of the action potential propagation modeled by the anisotropic Bidomain system coupled with the Luo-Rudy I membrane model. The results show a very good overall accuracy of ARI estimates of APD, with quite low mean discrepancies and standard deviations. The correlation coefficients between ARI and APDs are not as good as the ones of the associated repolarization time (RT) markers, but are always greater than 0.90 except for some homogeneous and ischemic slabs. While highly reliable repolarization sequences can be derived from extracellular RT markers, ARI spatial distributions are not always locally accurate and well correlated estimates of the associated APD spatial distributions. In particular, EG-based ARIs can fail near the borders of the ischemic region (e.g. in presence of linear ST ramp or in absence of a T wave), and in such cases HMAP-based ARIs may offer reliable alternatives for estimating APDs.