Depth-Sensitive Algorithm to Localize Sources Using Minimum Norm Estimations

The main objective of this paper is to apply and evaluate the neural source localisation accuracy of a new depth-sensitive algorithm using physiological data. This new algorithm is based on the behaviour of the dispersion of the minimum norm solutions (MNE), and was already tested with simulated data [1], yielding an accuracy of 2 to 4 mm in noise free situations, and a mean accuracy of 10 mm for more disadvantageous situations. We estimate now the neural source depth in EEG recordings, namely in focal epileptic interictal paroxisms and in N100 auditory evoked potentials of normal volunteers. We show that the accuracy of the method is comparable with the commonly used dipolar localizations, when it was applied to those bio signal recordings. It was revealed that, under adequate constraints, the algorithm is suitable for the estimation of the depth of one or two simultaneous neural generators, using a rather simple MNE approach. This study, demonstrating that MNE can handle spatial-limited sources successfully, opens the possibility to localize both quasi-punctual and extended neural generators using only the simplest MNE algorithm.