Human Brain Mapping Abstract (June 9-15, 2006)
Co-modulation of spectral power between maximally
independent brain sources
Julie Onton and Scott Makeig
EEG data recorded at scalp
electrodes is highly correlated between channels. This phenomenon is due to
single sources projecting to multiple electrodes, as well as some smearing of
the electric field by the skull and scalp. Using independent component analysis
(ICA), these
mixed signals can be separated into EEG sources whose locations in the brain
can be approximated. The resulting signals are mostly uncorrelated and the
scalp projections unique. As a novel
exploration of the spectral modulation of these signal over time, we decomposed
overlapping 1-sec epochs from continuous data collected during an emotional
imagery task using a fourier
transform. These data were then submitted to ICA which returned, for each subject, 15
spectral modulation factors or patterns, which included a power spectrum
template for each IC included in the analysis. These spectral co-modulation
factors revealed 4 major patterns that were reasonably preserved across
subjects: 1) alpha and harmonic modulation at the peak frequency, 2) a
shift up in alpha peak frequency, 3) a shift down in alpha peak
frequency and 4) a broad shift in 20-40 Hz activity. The alpha modulations and
co-modulations were largely posterior, with peak frequency modulations being
mostly confined to the occipito-parietal junction.
The alpha shifts occurred in occipito-parietal as
well as somato-motor regions. High frequency power
shifts were the most widespread with extensive co-modulation between all major
regions of the brain including occipito-parietal, somato-motor and frontal areas. The results indicate that
independent brain sources
modulate oscillatory power and frequency in non-phase-locked
unison with other components, suggesting an influence of modulator
neurotransmitters projecting to specific combinations of brain regions
simultaneously. The release of particular neurotransmitters in conjunction with
current synaptic input, and likely myriad other factors, may cause predictable
shifts in local field oscillations, thus drawing physically separated regions
of the brain into relative unison. The advantage of spectral co-modulation can
only be speculated at this point, but may contribute to enhanced synaptic
communication between the co-modulated regions.