Accurate inverse calculation of ventricular depolarization isochrones requires inhomogeneous models, precise heart geometry, and around 64 leads, but is robust to normal ECG noise.
This paper describes experimental studies on the influence of inhomogeneities in the volume conductor, and model errors and measurement errors on the inversely calculated isochrones of ventricular depolarization. The results indicate that in the inverse problem the use of inhomogeneous models is essential, that knowledge of the exact heart geometry (shape and orientation) is important, and that signal error (noise) as encountered in normal ECG recordings is not critical. The number of (measurement) leads required is found to be on the order of 64.
Cuppen et al. (Mon,) studied this question.