Charting Normative Ranges of EEG Oscillations Across Childhood and Adolescence
Poster No:
981
Submission Type:
Abstract Submission
Authors:
Ymke Verduyn1, Mohammad Zamanzadeh Nasrabadi1, Augustijn de Boer2, Tomas Ros3, Thomas Wolfers4, Richard Dinga1, Marie Šafář1, Marijn van Wingerden1, Andre Marquand2, Seyed Kia1
Institutions:
1Tilburg University, Tilburg, Brabant, 2Donders Institute for Brain, Cognition and Behaviour, Nijmegen, Gelderland, 3University of Geneva, Geneva, Canton of Geneva, 4University of Tübingen, Tübingen, Baden-Württemberg
First Author:
Co-Author(s):
Introduction:
Normative modeling provides a statistical framework for decoding heterogeneity in large-scale brain imaging cohorts. In this study, we extend the application of normative modeling to electroencephalography (EEG) data to establish normative ranges for oscillatory brain activity. Building on the work of Tröndle et al. (2022), which employed multivariate Bayesian generalized linear models to analyze aperiodic and oscillatory components of alpha activity, our study makes four key contributions: (1) normative ranges are derived for four canonical frequency bands−theta, alpha, beta, and gamma; (2) the analysis is restricted to individuals without clinical diagnoses to ensure a better reference for the general population; (3) a more flexible model is employed to account for both non-linearity and heteroscedasticity in the data; and (4) variation percentiles are estimated, providing a detailed characterization of normal variability within the population.
Methods:
We used closed-eyes resting-state EEG recordings of 230 undiagnosed individuals aged 5–21 years from the Healthy Brain Network dataset (Alexander et al., 2017). Data was collected across three sites using a 128-channel EEG Geodesic HydroCel system. An automated preprocessing pipeline was applied to the raw EEG signals, including bandpass filtering, re-referencing to the average, and artifact removal via automatic Independent Component Analysis (ICA) for ECG and EOG artifacts (Li et al., 2022). Subsequently, the FOOOF (Fitting Oscillations and One Over f) algorithm (Donoghue et al., 2020) was used to parameterize the neural power spectra into aperiodic and periodic components. The power spectra were adjusted to account for aperiodic activity, and relative power for the four canonical frequency bands (theta, alpha, beta, and gamma) averaged across all EEG channels was extracted from both the original and adjusted spectra. We then used Hierarchical Bayesian Regression (HBR) (de Boer et al., 2024; Kia et al., 2020) from the PCNToolkit (Rutherford et al., 2022) to estimate the normative model on the training set (50% of samples). HBR offers a flexible framework to model multi-site, non-linear, non-Gaussian, and heteroscedastic data.
Results:
Figure 1 shows the estimated normative ranges for relative spectral powers across four frequency bands (rows) before and after adjustment for the aperiodic component of the signal (columns). While the general trends are similar for the original and the adjusted power spectra, removing non-oscillatory components results in differences in the estimated centiles and interpersonal variability. Note that given the limited samples after the age of 14, the accuracy of estimated centiles declines, therefore, should be interpreted with caution. For the theta band, a decreasing trend across age can be seen. The alpha frequency shows an increasing trend, consistent with Tröndle et al. (2022). The beta original spectrum indicates an initial decrease up to the age of eight, followed by an increase, while for the adjusted spectrum a general upward trend is observed. Finally, the gamma band shows a downward trend in early childhood, followed by an increase starting around the age of ten.
Conclusions:
We derived normative ranges of neural oscillations from EEG data spanning childhood and adolescence, focusing exclusively on participants without clinical diagnoses to ensure an accurate representation of the general population. Our findings highlight the impact of adjusting the power spectrum on the estimated variability of relative power in the oscillatory components. These results underscore the importance of developing normative models using larger datasets that encompass the full human lifespan for a more comprehensive understanding of brain activity across developmental stages.
Lifespan Development:
Aging
Early life, Adolescence, Aging 1
Modeling and Analysis Methods:
Bayesian Modeling
EEG/MEG Modeling and Analysis 2
Keywords:
Aging
Data analysis
Electroencephaolography (EEG)
Machine Learning
Modeling
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
de Boer, A. A. (2024). Non-gaussian normative modelling with hierarchical bayesian regression. Imaging Neuroscience, 2, 1–36.
Donoghue, T. (2020). Parameterizing neural power spectra into periodic and aperiodic components. Nature Neuroscience, 23(12), 1655–1665.
Kia, S. M. (2020). Hierarchical bayesian regression for multi-site normative modeling of neuroimaging data. Medical Image Computing and Computer Assisted Intervention – MICCAI 2020, Lecture Notes in Computer Science, 12267. Springer, Cham.
Li, A. (2022). Mne-icalabel: Automatically annotating ica components with icalabel in python. Journal of Open Source Software, 7(76), 4484.
Rutherford, S. (2022). The normative modeling framework for computational psychiatry. Nature Protocols, 17(7), 1711–1734.
Tröndle, M. (2022). Decomposing the role of alpha oscillations during brain maturation. eLife, 11, e77571.