EEG preprocessing is required to attenuate artifacts and improve signal quality for downstream analysis.
This MNE preprocessing repository is implemented using the MNE-Python package.
The script rs_eeg_prep_for_psd_analysis is designed for resting-state EEG preprocessing, specifically for power spectrum–based analyses, including 1/f (aperiodic) and oscillatory (periodic) components.
- Downsampled to 1000 Hz
- Band-pass filtered between 0.1–45 Hz or 0.1–95 Hz using FIR filtering, followed by notch filtering at 50 Hz and its harmonic (100 Hz)
- Visually inspected to identify noisy segments (e.g., muscle artifacts affecting multiple channels), which were manually labeled as
bad_artifact - Bad channels (high-amplitude artifacts consistent across recordings) were removed and spherically interpolated
- Data were re-referenced to the average reference
- To enhance spatial specificity and reduce volume conduction at the sensor level, a surface Laplacian (current source density, CSD) transform was applied with parameters (Stiffness= 4, Legendre polynomial= 80, Regularization parameter= 10⁻³)
- Independent Component Analysis (ICA) using the FastICA algorithm was applied to identify and remove non-brain artifacts (e.g., eye blinks and muscle activity)
-
Bertino, S., Ghaderi-Kangavari, A., Meder, D., Vinding, M.C., Raaf, N., Christiansen, L., Thomsen, B.L.C., Løkkegaard, A., Quartarone, A., Beck, M.M. and Siebner, H.R. (2026). Increased aperiodic offset and heightened alpha power characterize resting-state EEG activity in Parkinson′ s disease. medRxiv, 2026-01. doi: https://doi.org/10.64898/2026.01.20.26343832
-
Ghaderi-Kangavari, A., Rad, J. A., Parand, K., & Nunez, M. D. (2022). Neuro-cognitive models of single-trial EEG measures describe latent effects of spatial attention during perceptual decision making, Journal of Mathematical Psychology, 111. doi: https://doi.org/10.1016/j.jmp.2022.102725