Neuromag software

Epileptiform events were identified according to the glossary of terms most commonly used by clinical electroencephalographers and recommendations The American Clinical MEG Society (ACMEGS) [4 (link),28 (link)] (Kane et al. 2017; Bagić et al. 2011). The epileptiform discharges were reconstructed at the peak using the multi-dipole modeling procedure (ECD, Equivalent current dipole) implemented into the Elekta Neuromag software, using a spherical head model. Three expert reviewers agreed on the marked events (T.S, A.K., A.Kr.). It is important to stress that not all spikes were visually marked, but only the amount necessary to compile a patient report.

MEG data preprocessing was performed according to the standard practice for clinical MEG research [23 (link),24 (link)]. The data were processed by МахFilter (Elekta Neuromag Software) using temporal signal space separation (tSSS, [25 (link)]). For each patient, out of the 1–2 hours available recording, we selected 20 minutes with the highest number of visually marked spikes during sleep. MEG data were visually inspected and noisy segments were excluded for the automated spike detection pipeline (< 2 minutes out of 20-minutes recording). Eye blinks and heart beats were projected out by ICA decomposition.

MEG data were recorded from a 306-channel whole-head MEG system (Elekta, Helsinki, Finland) and source localization analysis was performed using the vendor NeuroMag software (Elekta, Helsinki, Finland). Individual spike analysis was performed on data segments containing visually identified epileptiform discharges (9 (link)). The location, orientation and strength of dipole sources that best fit the measured magnetic fields were calculated typically using single equivalent current dipole model at the peak of the global field power of each interictal activity (10 (link), 11 (link)). The final results were represented by one or several clusters of dipoles superimposed on the patient's coregistered 3D T1-weighted MPRAGE. We exported the location of the MEG dipoles by printing them as high-intensity points on top of the MRI (dipole head 60%, MRI slices 5 mm, separation at print output 1 mm). The MRI DICOM images after the printing process were then imported and coregistered to the base image volume (Figure 1). The high-intensity points representing the dipole locations were then segmented and their 3D coordinates were saved as a list of localize points, and visualized in conjunction with all the other data.