Abstracts

Rationale: Planning for dominant hemisphere epilepsy surgery often requires functional language mapping for surgical planning and patients frequently report lanugage deficits despite standard mapping procedures, suggesting the need for more sensitive mapping techniques. Visual object naming is most often tested, however, no standard language battery is currently administered. Subdural electrodes are implanted for seizure localization, and task-related increases in electrocorticographic (ECoG) high gamma (65-150 Hz) activity recorded from these electrodes during language task performance has been used to detect cortical activation with excellent spatial and temporal resolution. We hypothesized that ECoG mapping would discriminate the cortical representations of auditory descriptive naming (AN) and visual object naming (VN), and that areas activated during AN, but not during VN, would lie within the boundaries of standard dominant resections. Methods: We studied 16 epilepsy patients (mean age 32.6, range 14-50) in whom subdural electrodes were implanted over dominant cortex for epilepsy surgery planning. Patients underwent ECoG mapping of visual object naming and auditory descriptive naming. We performed time-frequency analyses of ECoG signals using a matching pursuit algorithm, and detected statistically significant increases in event-related high gamma activity with respect to a pre-stimulus baseline (defined as ECoG+ sites). ECoG+ sites were localized by co-registration of post-surgical CTs to pre-surgical volumetric MRIs and normalized onto a Montreal Neurologic Institute brain atlas, then assigned appropriate Brodmann areas. The number of AN and VN ECoG+ sites were compared in anterior temporal (3.5-4.5 cm posterior to the dominant temporal tip) and frontal (anterior to the primary motor cortex, excluding BA 44 and 45) regions of interest, that represented the boundaries of potential dominant temporal and frontal resections. Results: Of the 932 total electrodes tested, 230 were ECoG+, 162 during AN (97 during AN only), 133 during VN (68 during VN only), and of these, 65 during both AN and VN. AN-specific regions were identified in all patients, mainly over dominant anterior temporal and frontal cortices, and nearly half (29) of VN-specific regions were identified over dominant basal temporal-occipital cortex. Over half (35) of sites that were ECoG+ during both tasks localized to perisylvian, primary motor (BA4) or somatosensory (BA1-3) cortices. A significantly larger number of AN than VN electrodes were located within the boundaries of a potential dominant anterior temporal or frontal lobectomy. Conclusions: ECoG spectral mapping in a relatively large sample of patients revealed distinct and shared ECoG+ regions during auditory descriptive naming and visual object naming. AN sites were more likely than VN sites to be located within the potential boundaries of dominant anterior temporal or frontal lobe resections, suggesting that functional anatomy supporting AN may be in jeopardy when planning dominant resections on the basis of VN mapping alone.