.png){kind=icon}
.png){kind=icon}
Return
Selecting methods for a modular EEG pre-processing pipeline: An objective comparison
.png){kind=icon}
OA
AI
DOI：10.1016/j.bspc.2023.105830.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
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 of the pre-processing on the signal is admitted and accepted, there is an increasing effort to better understand to which extent such an influence may affect the analysis results, and which may be the best practices for the correct data pre-processing.Pre-processing procedures include different steps, and each of them may induce modifications affecting the study results. Thus, we analyze the effect of different methodologies at each step and propose quantitative parameters for the choice of the preferable strategy.We illustrate how method selection may affect the quality of EEG signal in an Action Observation and Motor Imagery protocol, using quantitative indices. We analyzed the effect of different strategies for early-stage data preparation; two independent component analysis (ICA) algorithms (SOBI and Extended Infomax) used for artifact removal; and four re-reference approaches (Common averaged reference-CAR, robust-CAR, reference electrode standardization technique - REST, and reference electrode standardization and interpolation technique - RESIT). The effects of the different pipelines were also evaluated through the computation of event-related spectral perturbation (ERSP) of the sensorimotor rhythm. Results showed that signal segmentation significantly affects the cleaning procedure, while comparable results are obtained across ICA approaches. Finally, similar topographical representations were obtained after the application of CAR, REST, and RESIT rereferencing approaches, where rCAR showed the most different ERSP topographical pattern.
Keywords:
Pre-processing
EEG
ICA
Re-reference
Event-related spectral perturbation
Signal processing
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.9
Papers:
9.8K
Citations:
2.4W