Selecting methods for a modular EEG pre-processing pipeline: An objective comparison

• DOI: 10.1016/j.bspc.2023.105830
• Publication Date: 2023-12-14T17:08:25Z
• Authors: Stefania Coelli; Alessandra Calcagno; Chiara Maria Cassani; Federico Temporiti; Pierluigi Reali; Roberto Gatti; Manuela Galli; Anna Maria Bianchi
• Keywords: Artifact (error)

Electroencephalography (EEG) to study brain functions has become fundamental in many research settings across very different protocols. Indeed, a plethora of processing methods have been developed, for both data preparation (pre-processing) and analysis. While an effect the pre-processing on signal is admitted accepted, there increasing effort better understand which extent such influence may affect analysis results, be best practices correct pre-processing. Pre-processing procedures include steps, each them induce modifications affecting results. Thus, we analyze methodologies at step propose quantitative parameters choice preferable strategy. We illustrate how method selection quality EEG Action Observation Motor Imagery protocol, using indices. analyzed strategies early-stage preparation; two independent component (ICA) algorithms (SOBI Extended Infomax) used artifact removal; four re-reference approaches (Common averaged reference-CAR, robust-CAR, reference electrode standardization technique – REST, interpolation RESIT). The effects pipelines were also evaluated through computation event-related spectral perturbation (ERSP) sensorimotor rhythm. Results showed that segmentation significantly affects cleaning procedure, while comparable results are obtained ICA approaches. Finally, similar topographical representations after application CAR, RESIT re-referencing approaches, where rCAR most ERSP pattern.