Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis. II. Event-related synchronization in the gamma band

It has been shown in animals that neuronal activity in the 'gamma band' (>30 Hz) is associated with cortical activation and may play a role in multi-regional and multi-modal integration of cortical processing. Studies of gamma activity in human scalp EEG have typically focused on event-related synchronization (ERS) in the 40 Hz band. To assess further the gamma band ERS further, as an index of cortical activation and as a tool for human functional brain mapping, we recorded subdural electrocorticographic (ECoG) signals in five clinical subjects while they performed visual-motor decision tasks designed to activate the representations of different body parts in sensorimotor cortex. ECoG spectral analysis utilized a mixed-effects analysis of variance model in which within-trial temporal dependencies were accounted for. Taking an exploratory approach, we studied gamma ERS in 10-Hz-wide bands (overlapping by 5 Hz) ranging from 30 to 100 Hz, and compared these findings with changes in the alpha (8-13 Hz) and beta (15-25 Hz) bands. Gamma ERS (observed in three out of subjects) occurred in two broad bands-'low gamma' included the 35-45 and 40-50 Hz bands, and 'high gamma' the 75-85, 80-90, 85-95 and 90-100 Hz bands. The temporal and spatial characteristics of low and high gamma ERS were distinct, suggesting relatively independent neurophysiological mechanisms. Low gamma ERS often began after onset of the motor response and was sustained through much of it, in parallel with event-related desynchronization (ERD) in the alpha band. High gamma ERS often began during, or slightly before, the motor response and was transient, ending well before completion of the motor response. These temporal differences in low and high gamma suggest different functional associations with motor performance. Compared with alpha and beta ERD, the topographical patterns of low and high gamma ERS were more discrete and somatotopically specific and only occurred over contralateral sensorimotor cortex during unilateral limb movements (alpha and beta ERD were also observed ipsilaterally). Maps of sensorimotor function inferred from gamma ERS were consistent with maps generated by cortical electrical stimulation for clinical purposes. In addition, different task conditions in one subject produced consistent differences in both motor response latencies and onset latency of gamma ERS, particularly high gamma ERS. Compared with alpha and beta ERD, the topography of gamma ERS is more consistent with traditional maps of sensorimotor functional anatomy. In addition, gamma ERS may provide complementary information about cortical neurophysiology that is useful for mapping brain function in humans.