[Eeglablist] Best Approach for Running ICA on EEG Data for Classification(ME & MI)

[Pragati Dode]

Hello EEGLAB Community,

I am a beginner in ICA and working with an EEG dataset of size (12,162,240
× 46) containing six different actions. Initially, I ran ICA on the entire
dataset and then separated the ICA components label-wise. However, I
noticed that the ICA components remained the same across all labels, likely
because the same unmixing matrix was applied.

To address this, I first separated my EEG data label-wise and then ran ICA
on each label individually. Now, I observe different ICA components for
each label.

I am working on classifying my data into motor execution and motor imagery.
Which approach would be more suitable for running ICA in this context?
Should I apply ICA to the entire dataset first or process each label
separately and then combine all ICA ?

Looking forward to your insights!

Best regards,

Pragati Dode.