Neuroimaging of Ictal MEG: An Evaluation of Magnetic Source Imaging Techniques

Magnetoencephalography (MEG) is a valuable tool in the presurgical workup of refractory epilepsy patients. Ictal Magnetic Source Imaging (MSI) can more accurately localize the ictal onset zone, aiding presurgical planning. Nevertheless, the optimal approach for ictal MSI remains undetermined. To evaluate the effectiveness of distinct ictal MSI techniques, assessing their performance based on the ictal onset pattern (IOP). Design: Retrospective study. 16 ictal MEG events from 12 epilepsy patients were retrospectively analyzed. Techniques employed include the traditional sECD, and alternative approaches comprising the linearly constrained minimum variance (LCMV) beamforming, kurtosis beamforming, and dynamic statistical parametric mapping (dSPM). Seizures were classified into IOP groups: ictal discharge, rhythmic activity (RA), slow RA, and fast activity. Sublobar and lobar concordance and the minimum Euclidean distance (Dmin) were evaluated using SEEG data as ground truth. sECD fitting failed for three seizures, whereas alternative techniques demonstrated superiority. LCMV showed the highest sublobar concordance. No significant differences in Dmin across techniques were found. All techniques performed better in the ictal discharge group. Performance declined in the rhythmic activity IOP group, especially in lower frequencies, although LCMV performed better. ECD, beamforming, and dSPM are effective techniques for ictal MEG analysis. Beamforming techniques are particularly important when ECD is unsuitable. The IOP should be considered when selecting the appropriate ictal MSI technique. Optimizing MSI techniques and customizing them based on seizure characteristics can aid in invasive study planning and potentially improve post-surgical outcomes.