Abstracts

Rationale: Normal developmental changes in language lateralization and localization have been well documented using non-invasive neuroimaging. However, children with intractable epilepsy may have altered language network development, due to pathology and/or seizure activity. Synchronized gamma oscillations, particularly high gamma activity (HGA; >60Hz) was suggested to play a role in binding cortical components of the language network in recent electrocorticography (ECoG) studies. Magnetoencephalography (MEG) has been successfully applied to map language in a clinical context, by localizing event-related synchrony or desynchrony (ERS or ERD, respectively), particularly lower frequencies. ECoG has revealed spontaneous responses in gamma frequencies higher than those previously observed in non-invasive recordings in contrast to a majority of MEG studies examining lower frequency oscillatory changes. The purpose of this study is, 1) to characterize the HGA network for language in typical development, and 2) its pattern in children with epilepsy. Methods: MEG recordings were conducted on a 151 channel CTF magnetometer for ten healthy right-handed subjects (15-18 years old, 5 males). A verb generation task with objects whose names are familiar to typically-developing 5 year old children was applied. All trials were averaged based on the time of presentation of pictures as a trigger. Source localization was evaluated based on a current density reconstruction algorithm (sLORETA) in four different bandwidths: alpha (8-12Hz), beta (13-30Hz), gamma (30-70Hz) high gamma (70-150Hz) applied for five time windows: 0-200ms, 100-300ms, 200-400ms, 300-500ms and 400-600ms following picture presentation. The source coherence analysis was then applied for each five time window with 4 different bandwidths. The localization of a language area was defined as a lag time is less than 10ms within 2 voxel (30 mm²) and its connectivity was defined as the lag time is more than 30ms between different regions of interests (ROIs). Results: Source localization using the sLORETA demonstrated consistent localization with high gamma activity (80-150Hz) preferentially in in the left inferior frontal region in eight subjects during the time ranges of 200-400ms and300-500ms after the picture presentation. The source coherence model showed a 20-30ms phase shift from visual cortex to left inferior frontal cortex reflecting sequential processing of picture stimuli. In the remaining two participants, one showed right inferior frontal activation and the other had bilateral inferior frontal activations with bilateral synchronization in time range of 300-500ms and 400-600ms. Conclusions: The functional response properties of high gamma activity (HGA) are distinct. This method may advance the non-invasive presurgical language localization with detection of a more specific localized cortical area before the surgery. The HGA findings are consistent with its proposed binding role in models of neural computation.