Poster #3 - Assessing electroencephalography (EEG) power spectra using a novel an algorithmic approach in early childhood

• In Event: Poster Session 07

Abstract

Electroencephalography (EEG) is a psychophysiological measure of brain activity that has increasingly been used in research with infants and children to better understand early cognitive and emotional development (Brito et al., 2016). Traditionally, studies examining oscillations in EEG electrical activity utilize spectral analysis, which involves decomposing the neural signal into predetermined frequency bands in the power spectrum (e.g., delta (1–3 Hz), theta (4–7 Hz), alpha (8–12 Hz)) in adults. However those ranges are known to vary as a function of age (Saby & Marshall, 2012). In recent years, there has been growing recognition among researchers that analyses using fixed boundaries for the frequency bands may not be capturing all of the potential information from the distinct features of the EEG signal (Pathania et al., 2021; Robertson et al., 2019).

Fitting of oscillations and one-over f noise (FOOOF) is an algorithmic approach to EEG data analysis that utilizes aperiodic, or non-oscillatory, background activity from the EEG power spectrum. The aperiodic or background component of EEG reflects the exponential decrease of power at higher frequencies, and the rate of that decay of power — or the underlying slope of the power spectrum — is referred to as 1/f noise (Donoghue et al., 2020; Pathania et al., 2021). However, recent studies have found evidence suggesting that the 1/f component should not be treated as noise, but rather as a meaningful signal (Haller et al., 2018). Current work has also shown that features of the oscillatory (periodic) and aperiodic signal change across the lifespan, and researchers posit that understanding age-related changes in these neural signals may provide greater insight into what psychophysiological processes are occurring during sensitive developmental periods (Hill et al., 2022).

Aim: The goal of this project is to further study and replicate research on the utility of aperiodic signal components in developmental EEG research. In an early childhood sample (n = 92, 51.1% female, age range: 4-8 years, average age = 5.8 years), we will explore associations between age and aperiodic features of the EEG power spectrum.

Method: EEG data were obtained using a 128 electrode HydroCel Sensor Net System with Net Station software. Eyes open resting-state data were collected for 7 minutes while children viewed a fixation cross.

Analysis: EEG data was preprocessed and cleaned using independent component analysis. FOOOF tools will be used to parameterize (i.e., to separate the periodic and aperiodic EEG signal) the data and help measure spectral slope and timing of individual alpha peaks. Given prior research we expect that the timing of the alpha peak will be positively associated with age where older children will exhibit peak alpha at higher frequencies. We also expect that slope will decrease with age as previous work suggests that higher frequencies contribute more to the overall signal with increasing age across childhood. Data for this investigation have been collected as a part of an ongoing study, and more details about methods and future analysis plan have been pre-registered through the Open Science Foundation (osf.io/q3shc).

Authors

• Summer Motton, University of North Carolina at Chapel Hill
  Presenting Author

• Celina Meyer, University of North Carolina at Chapel Hill

• Madeline Mae Robertson, University of North Carolina at Chapel Hill

• Margaret A. Sheridan, University of North Carolina at Chapel Hill