Spectral and source structural development of mu and alpha rhythms from infancy through adulthood

Abstract

Objective: To assess the developmental trajectory of spectral, topographic, and source structural properties of functional mu desynchronization (characterized during voluntary reaching/grasping movement), and investigate its spectral/topographic relation to spontaneous EEG in the developing alpha band.

Methods: Event related desynchronization (ERD) and power spectral density spectra/topography are analyzed in 12 month-old infants, 4 year-old children, and adults. Age-matched head models derived from structural MRI are used to obtain current density reconstructions of mu desynchronization across the cortical surface.

Results: Infant/child EEG contains spectral peaks evident in both the upper and lower developing alpha band, and spectral/topographic properties of functionally identified mu rhythm strongly reflect those of upper alpha in all subject groups. Source reconstructions show distributed frontoparietal patterns of cortical mu desynchronization concentrated in specific central and parietal regions which are consistent across age groups.

Conclusions: Peak frequencies of mu desynchronization and spontaneous alpha band EEG increase with age, and characteristic mu topography/source-structure is evident in development at least as early as 12 months.

Significance: Results provide evidence for a cortically distributed functional mu network, with spontaneous activity measurable in the upper alpha band throughout development.

Keywords: EEG; Motor development; Mu rhythm.

Conflict of interest statement

Conflict of Interest

None of the authors have potential conflicts of interest to be disclosed.

Copyright © 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.

Figures

Figure 1

Experimental setup and participants. Subjects from three age groups corresponding to Adult, 4-year, and 12-month olds completed trials consisting of a three-second baseline interval followed by a grasp execution interval during which the subject reached towards and grasped a toy (A). During the baseline, adult and 4-year subjects viewed a white foam core board with a black shape or pattern (B), whereas 12-month subjects viewed a moving pinwheel. During grasp execution, a toy placed on a retractable table top was pushed toward the subject to within reaching distance (C).

Figure 2

Age-matched head models with co-registered electrode positions for each subject group, constructed using averaged T1-weighted anatomical templates from the University of South Carolina McCausland Brain Imaging Center Neurodevelopmental MRI Database. Boundary Element Method (BEM) forward models consist of four nested surfaces corresponding to scalp, skull, intracranial space, and gray matter, which were extracted using Freesurfer and MNE software. Forward models were computed using openMEEG, with conductivity values of 0.33, 1.67, 0.0042, and 0.33 for gray matter, intracranial space, skull, and scalp, respectively.

Figure 3

Log power spectral density by frequency from 1 – 20 Hz for each subject group (A, C, E), along with peak histograms (B, D, F) depicting the variability across bootstrap samples in frequencies at which alpha spectral peaks (between 4 – 13 Hz) were found. Average log PSD across subjects is indicated by red bars, whereas semi-transparent light red bars indicate the 95th percentile of log PSD across bootstrap samples. For 12-month olds, bootstrap sample peaks (B) were tightly clustered in two distinct bands with means of 4.49 and 7.39 Hz corresponding to peaks observed in the mean spectrum (A). 4-year subjects also showed bootstrap peaks (D) which were tightly clustered with mean frequencies of 6.6 and 9.81 Hz corresponding to peaks observed in the mean spectrum (A). For adult subjects, the log PSD spectrum (E) didn’t show the same clear doubly-peaked structure. Instead, a single broad peak (with mean peak frequency of 10.51) spanning the classic adult alpha range captured 100% of the peak frequency spread observed across bootstrap samples (F).

Figure 4

Upper and lower alpha band topography for each subject group (A, C, E), along with the log ratio of normalized power in the upper/lower bands (B, D, F). Upper and lower alpha topographies (A, C, E) appear superficially similar within each subject group, showing large peaks over occipital electrodes (indicated by cyan), falling off to parietal. Adult subjects show an additional peak at frontal channels (indicated in blue) in both bands as well. The fine structure of topographic differences between these two bands is more readily observable comparing normalized log PSD topographies via log ratio (masked by statistical significance in B, D, F). For all subject groups, bilateral central electrodes show the largest relative contribution to the upper alpha band (C3/C4 electrode locations are indicated by green dots in B, D, E for comparison). For adults and 4-year olds (D, F), foci are tightly clustered about C3/C4, whereas the distribution for 12-month olds (E), while peaking at C3/C4, is more broadly distributed. The distinct bilateral central peaks evident in these maps have long been associated with mu rhythm topography.

Figure 5

Channel averaged ERD spectra for each age group (A, C, E, left plot) are indicated by blue bars for frequencies which met significance criterion (and gray bars otherwise), with black error bars indicating 95% bootstrapped confidence intervals. Also shown are histograms (A, C, E, right plot) depicting the variability across bootstrap samples in frequencies at which ERD peaks were found. For 12 month subjects (A) greater than 97.5% of ERD peaks across bootstrap samples were concentrated between 7 – 8 Hz. For 4-year subjects (C) 100% of ERD peaks were contained in the band from 8.5 – 10 Hz. Likewise for adults (E), 100% of ERD peaks were contained in the band from 10 – 12 Hz. For each age group, the band containing the vast majority of ERD peaks across bootstrap samples fell within the upper alpha band identified in figure 3. ERD topography for these designated mu bands of interest is depicted for each age group in B, D, F. Scalp maps show distinct bilateral peaks over central and anterior parietal areas (C3/C4 electrode locations indicated by green dots for comparison), similar to those observed in figure 4.

Figure 6

Cortical mu band desynchronization computed from sLoreta source reconstructions for adult subjects is shown on the reconstructed cortical surface (A), along with the same shown on the inflated surface in order to better visualize ERD in the cortical sulci (B). ERD is shown masked by significance in both maps, such that only data for cortical nodes with ERD significantly less than zero is visualized (adjusted empirical p

Figure 7

sLoreta source reconstructions of cortical mu band desynchronization depicted for 4-year subjects similar to figure 6. Similar to the pattern observed for adults, mu band ERD is greatest in the central and parietal cortices, which account for the majority of the top ranked cortical areas in both hemispheres.

Figure 8

sLoreta source reconstructions of cortical mu band desynchronization depicted for 12-month subjects similar to figure 6. A similar pattern to that observed for adults and four year olds is evident, with central and parietal cortices accounting for the majority of the top ranked cortical areas in both hemispheres. Specifically, the post-central, pre-central, para-central, and superior frontal areas fall in the top ten for both hemispheres of all subject groups. Similarly, the superior parietal, and supramarginal cortices, as well as the precuneus (all parietal regions) all fall into the top ten ranked areas for both hemispheres of all subject groups.