RIS ID
134954
Abstract
Background: The majority of electroencephalographic (EEG) investigations in normal ageing have determined EEG spectra from epochs recorded in the eyes-closed (EC) and/or eyes-open (EO) resting states, and summed amplitudes or power estimates within somewhat-arbitrary and/or inconsistently defined traditional frequency band limits. New method: Natural frequency components were sought using a data-driven frequency Principal Components Analysis (f-PCA) approach, optimised to reduce between-condition and between-group misallocation of variance. Frequency component correspondence was screened using the Congruence Coefficient and topographic correlations for potential matches on Condition and/or Group. The amplitudes of corresponding natural components were then explored as a function of these independent variables. Results: Separate f-PCAs with Young and Older adults' EC and EO data each yielded between six and nine components that peaked across the traditional delta to beta band ranges. Across these, two components were matched on Group and Condition, while a further six were matched on Condition (within-groups), and four on Group (within-conditions). Comparison with Existing Methods: Multiple frequency components were found within the traditional bands, and provided a wider perspective in terms of additional natural component details. In addition to novel insights, the well-documented age-related spectral reductions were seen in the common delta component, and in one EC (but no EO) alpha component. Conclusions: This combination of optimised f-PCA approach and component screening procedure has wide potential in the EEG field beyond the ageing topic explored here, being applicable across an extensive range of studies using EEG oscillations to explore aspects of cognitive processing and individual differences.
Publication Details
Barry, R. J., De Blasio, F. M. & Karamacoska, D. (2019). Data-driven derivation of natural EEG frequency components: An optimised example assessing resting EEG in healthy ageing. Journal of Neuroscience Methods, 321 1-11.