Abstracts

This study was undertaken to determine if it is possible to detect changes in scalp EEG that correspond with the behavioral measures of overt and silent object naming. Nine subjects with documented left temporal lobe epilepsy were selected for this study. All were surgical candidates. Language, in all cases, was shown by the intracarotid sodium amytal (Wada) test to be lateralized to the left hemisphere. EEG data (0.1-1000 bandwidth; 500 Hz sampling rate) was collected preoperatively during behavioral tasks. Dense array EEG recordings with 256 channels were utilized to improve spatial resolution. Subjects were presented with slides every 4 sec that contained images of 80 common objects. During one trial subjects named objects out loud and during a second trial objects were named silently. EEG data was analyzed in frequency bands that ranged from 25-50 Hz and were filtered with a Remez band-pass filter. Power spectral densities (PSDs) during baseline and object naming were calculated after obtaining Fourier transforms of the data. PSDs were averaged over the 80 data sets of each trial. Contour plot were constructed using the montage layout of the 256 channel array. In all cases, contour plots demonstrated that PSDs during both overt and silent naming were 2-3 times higher than the baseline values for all frequency bands. Maximal changes were localized to focal areas of the left cerebral hemisphere. These findings suggest that it may be possible to noninvasively lateralize, and eventually localize, cerebral regions essential for language. This may be accomplished by analysis of scalp EEG recordings that have optimal spatial resolution.