Human Brain Mapping
Florence, Italy
June 11-15, 2006
Julie Onton & Scott Makeig , Institute for Neural Computation, University of California San Diego, La Jolla CA
Independent co-modulation of spectral EEG power in subsets of independent brain sources
INTRODUCTION: Electroencephalographic (EEG) data recorded at nearby scalp electrodes are highly correlated, since single cortical sources project to nearly all the scalp electrodes, and the recorded field is smeared further by passing through the skull and scalp. Using infomax (or other) independent component analysis (ICA), the mixed scalp signals can be separated into maximally independent EEG sources whose brain locations can be approximated by standard inverse modeling methods. The resulting signals are nearly independent, and their scalp projections often can be fit to the projection of a single (or sometimes two bilateral) equivalent dipoles.
METHODS: To search for possible patterns of spectral co-modulation over time among maximally independent components, using a moving-window fourier transform we computed log spectral transforms of overlapping 1-sec epochs of the activation time series for independent components with dipolar maps from continuous 100-channel data collected during an emotion imagination task. These data were in turn decomposed by infomax ICA, returning for each subject 15 spectral co-modulation factors or templates, each comprised of a power spectral modulation pattern for each component included in the analysis.
RESULTS: Examination of the spectral co-modulation factors across data from 30 subjects revealed three major patterns: 1) co-modulations of alpha band power (and its harmonics) at the component peak alpha frequency; 2) small (circa 1-Hz) shifts, upwards or downward,s in the peak alpha frequency; and 3) broad co-modulations with a high-pass character (> 20 Hz). Each of these patterns had a distinct spectral character and affected, typically, only a few of the submitted components. Alpha band co-modulations were largely posterior. Peak frequency shifts were mainly confined to near the occipito-parietal junction, while co-modulations of alpha power occurred mainly within bilateral occipito-parietal and somato-motor component pairs. High-frequency co-modulations were the most spatially widespread, linking components with all cortical lobes including occipito-parietal, somato-motor and frontal areas.
DISCUSSION: These results indicate that the activity spectra of maximally independent brain sources, each the result of information-based spatial filtering of the scalp signals and each compatible with activity in one or sometimes two isolated cortical patches, exhibit spatially distinct and frequency-specific patterns of second-to-second spectral co-modulation. Many factors, including the actions of multiple neurotransmitter-labeled central arousal systems are known to produce changes in the power spectrum of local field oscillations, including broadband changes in gamma band power. Co-modulation of alpha band power, and coordinated alpha frequency shifts, may be involved in regulating synaptic communication between co-modulated regions or between their outputs and their inputs from more posterior and/or more frontal cortical areas. The presence of harmonically related peaks in the alpha power co-modulation templates suggests amplitude modulation of periodic but non-sinusoidal alpha-band activities.