Number of Current Dipoles in Source Localization from Neuromagnetic Fields

Simulation Study for Single and Two Dipole (s) Estimation

Abstract

In this paper, we have made a simulation study of localizing equivalent current dipole (ECD) sources from measured neuromagnetic fields. The purpose of this study is to make clear the problems caused by mis-estimation of the number of ECDs in the localization for single or two dipolar fields. When two-ECDs are localized for single dipolar fields, anomalous convergence results sometimes occur. In this case, the localized ECDs produce dipolar magnetic fields that exceed the measurement area. In the other results, one ECD is localized near the original dipole position and the other ECD position is scattered with initial guesses or noise patterns. In this case, localization results are characterized by some parameters, i. e., low contribution of magnetic fields produced by one ECD, short distance between the localized ECDs or large absolute value of correlation between magnetic fields produced by each ECD. When single ECD is localized for the fields produced by a main dipole and a noise dipole, location errors from the main dipole are more serious, where ECD is localized deeper than the main dipole position rather than shallower. In this case, goodness-of-fit value which is often used to judge the suitability of localization is relatively higher when the errors are large than the value when the errors are small. We have also studied the effects of the contributions of the magnetic fields produced by the target dipoles in two dipolar fields simulation. It was found that when the magnetic fields contribution of one of them is very low, single ECD method is more suitable to localize the main dipole than two-ECDs method, and when the contributions of both dipoles are high, two-ECDs localization is necessary.