Application of Hilbert transform to scalp EEG containing EMG
- Author(s): Freeman, Walter J, III
- Burke, Brian C
- Holmes, M D
- et al.
Objective: To evaluate rapid changes in regional EEG synchronization in normal subjects with spatial and temporal resolution exceeding prior art 10-fold.
Methods: A curvilinear array of 64 electrodes 3 mm apart extending 18.9 cm across the scalp was used to record EEG at 200/s. Analytic amplitude (AA) and phase (AP) were calculated at each time step for the 64 traces in the analog pass band of 0.5-120 Hz. AP differences approximated the AP derivative (instantaneous frequency). The AP from unfiltered EEG revealed no reproducible patterns. Filtering was necessary in the beta and gamma ranges according to a technique that optimized the correlation of the AP differences with the activity band pass filtered in the alpha range.
Results: The sizes of temporal AP differences were usually within ±0.5 radian from the average step corresponding to the center frequency of the pass band. Large AP differences were often synchronized over distances of 6 to 19 cm. An optimal pass band to detect and measure these recurring jumps in AP in the beta and gamma ranges was found by maximizing the alpha peak in the cospectrum of the correlation between unfiltered EEG and the band pass AP differences. Synchronized AP jumps recurred in clusters (CAP) at alpha rates in resting subjects and with EMG.
Conclusion: Cortex functions by serial changes in state. The Hilbert transform of EEG from high density arrays can visualize these state transitions with high temporal and spatial resolution and should be useful in relating EEG to cognition.