Abstracts

Rationale: Ictal recording of magnetoencephalography (MEG) may have better localizing value than interictal recordings in defining the ictal onset zone. However, conventional analysis using single-dipole models is difficult since early ictal discharges often have a low signal-to-noise ratio. Dynamic statistical parametric mapping (dSPM) is a new method for estimating brain activity. Several case studies have demonstrated the utility of dSPM for analyzing interictal epileptic discharges. In the present study, we compared dSPM with single-dipole modeling in the analysis of ictal MEG in patients with intractable epilepsy.Methods: We reviewed data from 5 patients with intractable partial epilepsy, who had ictal events during MEG recording. All patients underwent presurgical evaluations including long term scalp EEG monitoring, magnetic resonance images (MRI), positron emission tomography and single photon emission computed tomography at Massachusetts General Hospital (MGH) or Children’s Hospital. MEG tests were performed at Athinoula A. Martinos Center for Biomedical Imaging between 2005 and 2007. Simultaneous MEG and scalp EEG was recorded in each patient, using a 306 channel whole-head MEG system. After MEG recordings in 4 patients, high-resolution 3T MRI were acquired to support the MEG analysis. Previous MRI data was used for analysis in one patient. The MEG data were visually examined, and segments containing epileptic discharges were selected from the early part of ictal MEG for both single-dipole and dSPM analyses. Equivalent current dipoles (ECDs) were calculated at the peak of each ictal spikes using a single-dipole model. We also calculated dSPM to estimate the spatiotemporal cortical distribution of ictal spikes using a boundary element model (Neuron 2000; 26: 55-67). Both ECDs and dSPM solutions were mapped onto inflated images of cortical surface, which were derived from each patient’s MRI.Results: In each patient, 6-13 spikes were identified in the early phase of ictal MEG. Using the dSPM technique 70% of spikes in 4 patients and 100% of spikes in 1 patient mapped to the clinically determined epileptogenic zone at a lobar level. Other spikes showed widespread areas of activation including presumptive epileptogenic zones. By contrast, ECDs had poor statistical values such as goodness of fit and dipole moment in 80% (1 patient), 70% (2 patients), 60% (1 patient) and 30% (1 patient) of spikes. Furthermore, ECDs were located in an area inconsistent with the presumptive epileptogenic zones at a lobar level in 40% (2 patients), 30% (2 patients) and 10% (1patient) of spikes.Conclusions: In comparison with single-dipole methods, dSPM may be more useful for investigating ictal onset zones.