Frequency domain models of the EEG

Area of Science:

• Neuroscience
• Quantitative Electroencephalography (qEEG)
• Brain Signal Analysis

Background:

• The resting electroencephalography (EEG) cross-spectrum contains complex information about neural activity.
• Understanding the spatial structure of EEG signals is crucial for identifying brain generators.
• Previous analyses have not fully elucidated the distinct components within the EEG cross-spectrum.

Purpose of the Study:

• To analyze the structure of the normal resting EEG cross-spectrum (SVV(omega)) using complex multivariate statistics.
• To identify and characterize different types of neural generators contributing to the EEG signal.
• To apply a novel frequency-domain source estimation method.

Main Methods:

• Complex multivariate statistical analysis of resting EEG cross-spectra from 211 normal individuals (ages 5-97).
• Exploratory data analysis using Principal Component Analysis (PCA).
• Hypothesis testing, computer simulations, and application of a new frequency-domain dipole fitting method.

Main Results:

• The EEG cross-spectrum SVV(omega) decomposes into two main processes: an isotropic 'xi' process and localized processes.
• The 'xi' process exhibits spatial isotropicity, reflecting diffuse, radially symmetric cortical generators.
• Localized spectral peaks indicate spatially concentrated, correlated neural sources, explained by Spherical Harmonic Functions and dipole fitting.

Conclusions:

• The resting EEG cross-spectrum structure is characterized by both diffuse and localized neural generators.
• Spherical Harmonic Functions effectively describe the spatial patterns observed in EEG data.
• The new frequency-domain method aids in characterizing localized neural sources.