Variability of model-free and model-based quantitative measures of EEG

• Published in: arXiv.org 2018-01
• Main Authors: van Albada, Sacha Jennifer, Rennie, Christopher J, Robinson, Peter A
• Format: Article
• Language: English
• Subjects: Balances (scales); Brain; Continuum modeling; Frequencies; Frequency spectrum; Mathematical models; Parameter estimation; Peak frequency; Recording; Spectra; Thalamus; Time; Time lag
• ISSN: 2331-8422
• DOI: 10.48550/arxiv.1801.01711

The extent of intra-individual and inter-individual variability is an important factor in determining the statistical, and hence possibly clinical, significance of observed differences in the EEG. This study investigates the changes in classical quantitative EEG (qEEG) measures, as well as of parameters obtained by fitting frequency spectra to a continuum model of brain electrical activity. These parameters may have extra variability due to model selection and fitting. Besides estimating levels of intra-individual and inter-individual variability, we determine approximate time scales for change in qEEG measures and model parameters. This provides an estimate of the recording length needed to capture a given percentage of the total intra-individual variability. Also, if more precise time scales can be obtained in future, these may aid the characterization of physiological processes underlying various EEG measures. Heterogeneity of the subject group was constrained by testing only healthy males in a narrow age range (mean = 22.3 years, sd = 2.7). Resting eyes-closed EEGs of 32 subjects were recorded at weekly intervals over an approximately six-week period. Of these, 13 subjects had follow-up recordings spanning up to a year. QEEG measures, computed from Cz spectra, were powers in five frequency bands, alpha peak frequency, and spectral entropy. Of these, theta, alpha, and beta band powers were most reproducible. Of nine model parameters obtained by fitting model predictions to experiment, the most reproducible ones quantified the total power and the time delay between cortex and thalamus. About 95% of the maximum change in spectral parameters was reached within minutes of recording time, implying that repeat recordings are not necessary to capture the bulk of the variability in EEG spectra likely to occur in the resting eyes-closed state on the scale of a year.