Frontiers in Neuroscience (Sep 2022)
Source localization using virtual magnetoencephalography helmets: A simulation study toward a prior-based tailored scheme
Abstract
Magnetoencephalography (MEG) source estimation of brain electromagnetic fields is an ill-posed problem. A virtual MEG helmet (VMH), can be constructed by recording in different head positions and then transforming the multiple head-MEG coordinates into one head frame (i.e., as though the MEG helmet was moving while the head remained static). The constructed VMH has sensors placed in various distances and angles, thus improving the spatial sampling of neuromagnetic fields. VMH has been previously shown to increase total information in comparison to a standard MEG helmet. The aim of this study was to examine whether VMH can improve source estimation accuracy. To this end, controlled simulations were carried out, in which the source characteristics are predefined. A series of VMHs were constructed by applying two or three translations and rotations to a standard 248 channel MEG array. In each simulation, the magnetic field generated by 1 to 5 dipoles was forward projected, alongside noise components. The results of this study showed that at low noise levels (e.g., averaged data of similar signals), VMHs can significantly improve the accuracy of source estimations, compared to the standard MEG array. Moreover, when utilizing a priori information, tailoring the constructed VMHs to specific sets of postulated neuronal sources can further improve the accuracy. This is shown to be a robust and stable method, even for proximate locations. Overall, VMH may add significant precision to MEG source estimation, for research and clinical benefits, such as in challenging epilepsy cases, aiding in surgical design.
Keywords
