Hello Brainstorm team,
Following the tutorial on using ICA to remove noise in my high density EEG data, I repeatedly encounter with the issue shown below.
Could you please help me understanding my error?
Below I attach how am I computing ICA.
Thanks in advance
The issue is not with Picard nor with ICA. The trouble is that the channel used to sort the ICs (EEG REF) is constant. This triggered a unexpected error (now fixed in commit )
Commonly, artifactual signals (EOG, ECG) are used to sort the ICs, with the idea that the first ICs will be related to the influence of those artifactual signals on the brain recordings. I'm not sure what is the goal of using EEG REF to sort the ICs. If no signal is provided the sorting is done by the variance of the ICs.
Thank you for help, @Raymundo.Cassani ! This works
I misunderstood the " Sort components based on correlation with and thought that I could sort my ICA components based on the correlation with EEG REF.
Your understanding on the option is correct, when it is checked, ICs are descend-sorted according to their absolute correlation with the reference signal (EEG REF in your case).
The error was produced, because your EEG REF channel was flat. This was a bug, and it's now fixed.
