Abstracts

Rationale: A distributed source analysis of EEG and magnetoencephalography (MEG) provides a convenient spatiotemporal map of the onset and propagation of epileptic spikes. Source maps may be made for individual spikes, providing its time course in detail. Inspecting the distribution of multiple spikes in a patient is time consuming, and is difficult to display, whereas single dipole analysis provides dipole distribution of multiple spikes in a single image. Here, we propose a procedure to map the distribution of a population of epileptic spikes, and validate it with intracranial EEG (IEEG). Methods: We retrospectively analyzed MEG spikes in five patients with partial epilepsy who also underwent IEEG monitoring. L-2 minimum norm estimates (MNE) were computed from twenty-five interictal MEG spikes between -35 and 0 ms (=peak) by using a source space derived from each patient's MRI. For each spike, a threshold was determined by half of the maximum current magnitude at the peak. At each vertex, the number of spikes (0-25), which had a magnitude over the threshold, was obtained. We mapped these numbers at each vertex along with the time course, i.e., the maps reflect the probability of activation on the cortical locations within the group of spikes. For IEEG, we coregistered CT images acquired after implantation of intracranial electrodes onto cortical surface derived from MRI. Twenty-five IEEG spikes were selected in each patient and the threshold was determined in the same manner as the MEG analysis. Similarly to the MEG analysis, we counted the number of spikes activated in each electrode. For both MEG and IEEG, we determined the spike leading and peak involvement zones as the area activated in half of spikes in the earliest latency and peak, respectively. We visually inspected these zones on each individual's cortical surface, and determined whether the MEG-derived zones were consistent, partially consistent or inconsistent with the IEEG-derived zones at a sublobar level. Results: The spike leading zones were consistent between MEG and IEEG in one, and partially consistent in four of five patients, respectively. The peak involvement zones were consistent in one and partially consistent in four, respectively. Conclusions: We introduced a new mapping method that automatically displays the spatiotemporal source distribution from a spike population. The map of MEG spikes was consistent with IEEG findings.