Localisation of epileptic foci with electric, magnetic and combined electromagnetic models.

Abstract We compare the localisation of epileptic foci by means of (1) EEG, (2) magnetoencephalography (MEG) and (3) combined EEG/MEG data in a group of patients suffering from pharmaco-resistant focal epilepsy. Individual epileptic events were localised by means of a moving dipole model in a 4-shell spherical head approximation. A patient's epileptic activity was summarised by calculating the spatial density distribution (DD) of all localised events, and the centre of gravity of DD was considered the most likely locus of seizure generation. To verify these loci a subgroup of 6 patients was selected, in which seizures could be related to a clearly identifiable lesion in MRI. On average, the combined EEG/MEG approach resulted in the smallest error (1.8 cm distance between calculated locus and the nearest lesion border); using only MEG yielded the largest error (2.4 cm), while EEG resulted in an intermediate value (2.2 cm). In the individual patients, EEG/MEG would also rank intermediate, but never worst. In summary, combining EEG/MEG appears to be a more robust approach to localisation than using only EEG or only MEG. Finally, we also report on the use of the barbiturate methohexital as a safe method of increasing the number of spike events during an EEG/MEG recording session.