Abstracts

RATIONALE: Recent studies show that novel tools derived from nonlinear analysis of intracranial recordings permit the anticipation of an epileptic seizure by several minutes (J Martinerie et al., Nature Medicine 1998;4:1173-1176). Detection of dynamical pre-ictal changes has also been made possible from standard scalp electroencephalographic (EEG) signals. In order to directly compare the two patterns of pre-ictal changes, we undertook a study in patients with simultaneous depth and scalp EEG recordings. METHODS: We analyzed the EEG and the video recordings from 5 patients with medial temporal lobe epilepsy, using both scalp (international 10/20 system) and a combination of subdural and medio temporal intracerebral electrodes. Eight epochs of 60 minute duration, with 50 minutes before the seizure, were studied with a non linear measure of similarity that quantifies the extent to which the EEG dynamics differs between recordings taken at distant moments in time. A reference window of 300 seconds was determined between 40 and 50 minutes before the seizure. The significant deviations from the reference state of each contact, scalp and intracranial, are depicted as a function of time in spatio-temporal maps. RESULTS: There was a statistically significant correlation between the time course of scalp and intracranial pre-ictal changes, with comparable anticipation times, of the order of 400 seconds. From the scalp, the temporal lobe ipsilateral to the epileptic focus was involved in most cases, but pre-ictal changes had often a widespread distribution, projecting beyond the limits of the epileptogenic temporal lobe. CONCLUSIONS: These parallel results are further validating the accessibility to pre-ictal dynamical changes from standard, non invasive EEG. This opens the door to diverse clinical applications of these seizure anticipation strategies, such as developing patient warning systems, or facilitating the ictal-SPECT procedure.