[Eeglablist] Fix ICA rank deficiency in script

Krogmeier, Claudia M ckrogmei at purdue.edu
Mon Apr 19 07:24:14 PDT 2021

Dear list,

I have seen past eeglablist posts concerning how to fix ICA rank deficiencies, by changing to PCA parameter in the pop_runica function to the appropriate number of components for ICA, with

%EEG = pop_runica(EEG, 'icatype', 'runica', 'extended',1,'interrupt','on','pca',numberComps);

with numberComps as the appropriate # to use, especially if EEG is averaged referenced or chans are interpolated.

However, I have read on the EEGLAB tutorial page that: "If the rank of the data is lower than the number of channels, the EEGLAB pop_runica() function should detect it and uses different strategies to do so."

I just wanted to confirm that pop_runica will automatically adjust the components possible to detect, based on the dataset's rank, as suggested by the EEGLAB wiki.  Other EEGLAB list posts have led me to believe that it may be necessary for the user to specify numberComps, rather than let pop_runica automatically handle a rank deficiency issue. I am using an automated script.

Thank you very much,

More information about the eeglablist mailing list