Localization accuracy and temporal resolution of MEG: A phantom experiment

.Recent phantom studies have demonstrated localization accuracy of 1.4–3 mm for single trial recordings of dipole positions representing the somatosensory area, and of 14 mm for deeper locations. The duration of the dipole activation in these studies was too long and the stimulus was too strong to validate MEG's accuracy in localizing weak spikes of high frequency. It was recently demonstrated that such “MEG spikes” are widely distributed across the cortex, cerebellum and brainstem. The present study demonstrates MEG's ability to localize weak, transient brain activation in both single trials and averaged data. MEG data were collected using the CTF whole head system. A saline-filled spherical phantom was used with a dipole placed in one of three different locations. The dipole was activated with a train of 360 spikes. The data were analyzed using an Equivalent Current Dipole (ECD) and the Synthetic Aperture Magnetometry (SAM). On the cortex, dipole analysis fitted single trial data with localization error of 3.6 and 15 mm for dipole strengths of 80 and 9 nA-m, respectively. The averaged data for the same superficial locations were localized with accuracy between 1 and 2 mm. For deeper dipoles acceptable localization was only obtained for the averaged data. SAM analysis produced slightly better results than ECD. These results demonstrate that dipole fitting of average MEG data and SAM analysis can localize focal sources with an accuracy of 2 mm or better in the human cortex, and about 5 mm in deeper areas.