Estimating myocardial activation times by maximum likelihood estimation

In a cardiac isochronal map, myocardial dynamics are represented by activation times. Traditionally, ad-hoc methods (typically using the local minimum derivative of a unipolar electrogram, i.e., minimum first derivative algorithm) are used to detect myocardial activation times. We propose a statistical method, maximum likelihood (ML) estimation, to estimate myocardial activation times based on a dipole volume conductor model of a myocardial aggregate. Performance of the ML method is evaluated by repeated simulations with white noise. It is demonstrated that the ML method is more robust than the minimum first derivative algorithm. Further studies of the method are warranted perhaps with a more complicated model of noise and volume conductor, and with multiple electrodes.