ERPLAB Studio Panels: Artifact Correction with ICA - ucdavis/erplab GitHub Wiki

ICA-based artifact correction is implemented in ERPLAB Studio using the EEGLAB ICA panel in the EEG tab.

In its simplest form, ICA-based artifact correction involves three steps.

• Go to the EEGLAB ICA panel and click Decompose data to obtain the independent components (ICs) for the selected EEGset. This creates a new EEGset that contains the ICs along with the original EEG data.
• Determine which ICs reflect artifacts. This can be accomplished manually by examining the scalp topography and frequency content of each IC (by clicking the Inspect/label ICs button) and by examining how well the time course of a given IC matches the time course of the artifact in the original data (by selecting both Display chans and Display ICs in the Plot Settings panel). Alternatively, automatic classification can be performed by clicking the Classify IC by ICLabel button (which uses the ICLabel algorithm). If you use automatic classification, we strongly recommend visually confirming the classification by examining the scalp topography and time course of each IC.
• Go to the EEGLAB ICA panel and click Remove ICs to remove the ICs corresponding to artifacts. This creates a new EEGset that contains the corrected EEG data.

ICA will work much better if the decomposition process is performed on data that have been preprocessed to minimize idiosyncratic artifacts that do not have a fixed scalp topography. This is described in detail in Chapter 9 of Applied Event-Related Potential Data Analysis.

The Artifact Correction section of the ERPLAB Studio Tutorial provides a detailed example of how to implement artifact correction, including the optimal preprocessing.

You can also read the section on Artifact Correction in the EEGLAB Wiki for additional information about the EEGLAB artifact correction routines that are launched from ERPLAB.