Computer enhancement of scalp-sphenoidal ictal EEG in patients with complex partial seizures

Color topographic maps of scalp/sphenoidal ictal EEG records were computer-generated in 5 patients with medically refractory complex partial seizures of suspected temporal lobe origin. Seven ictal EEG records were analyzed by parsing them into a series of 1.28 sec epochs. User-interactive, computer techniques were utilized to replace eye movement and other artifactual segments in the peri-ictal records with nearby uncontaminated EEG segments. Artifact replacement techniques were designed to minimize or eliminate EEG discontinuities in those epochs in which artifact was removed. Significance probability maps consisting of z scores were constructed by comparing theta band power for each 1.28 sec epoch composing the peri-ictal period with the mean and standard deviation of theta power computed from a pre-ictal baseline period consisting of 50-75 epochs. The resulting maps were compared with available independent clinical measures in order to validate their usefulness. The independent measures consisted of non-invasive data (visual impressions of raw scalp/sphenoidal ictal EEGs and positron emission tomography) and invasive data (depth ictal recordings, pathological findings in resected temporal lobes, and surgical outcome). Ictal topographic maps appeared to either confirm or provide new localizing information in all 5 patients. In 3 of 5 patients, several seconds of localized theta suppression preceded localized theta augmentation during the peri-ictal period. The transition between the 2 states was very rapid (approximately 2 sec). The spatial locus of suppression was always in agreement with the spatial locus of augmentation. In general, the maps were considerably easier to interpret than the raw scalp/sphenoidal ictal EEGs: (1) Localized theta suppression in the maps was not evident from visual inspection of the raw ictal scalp/sphenoidal EEG. (2) Clear localized theta augmentation in the maps occurred well before the appearance of unambiguous phase reversals in the raw scalp/sphenoidal ictal EEG. (3) Mapping data were correctly lateralizing in 1 patient in whom visual interpretation of the raw scalp/sphenoidal ictal EEGs was considered non-localizing but often suggestive of a seizure origin in the hemisphere contralateral to that implicated by depth recordings, PET, and pathological data. These results suggest that topographic mapping of scalp/sphenoidal ictal EEGs in patients with complex partial seizures may eventually prove to be a useful adjunct to the interpretation of raw ictal recordings.