The influence of forward model conductivities on EEG/MEG source reconstruction

Abstract

In order to reconstruct the neuronal activity underlying measured EEG and MEG data both the forward problem (computing the electromagnetic field due to given sources) and the inverse problem (finding the best fitting sources to explain given data) have to be solved. The forward problem involves a model with the conductivities of the head, which can be as simple as a homogeneously conducting sphere or as complex as a finite element model consisting of millions of elements, each with a different anisotropic conductivity tensor. The question is addressed how complex the employed forward model should be, and, more specifically, the influence of anisotropic volume conduction is evaluated. For this purpose high resolution finite element models of the rabbit and the human head are employed in combination with individual conductivity tensors to quantify the influence of white matter anisotropy on the solution of the forward and inverse problem in EEG and MEG. Although the current state of the art in the analysis of this influence of brain tissue anisotropy on source reconstruction does not yet allow a final conclusion, the results available indicate that the expected average source localization error due to anisotropic white matter conductivity might be within the principal accuracy limits of current inverse procedures. However, in some percent of the cases a considerably larger localization error might occur. In contrast, dipole orientation and dipole strength estimation are influenced significantly by anisotropy. In conclusion, models taking into account tissue anisotropy information are expected to improve source estimation procedures.