[Eeglablist] ASR doesn't work on high density of electrodes for me

Rueda Parra, Sebastian (rued7682@vandals.uidaho.edu) rued7682 at vandals.uidaho.edu
Fri Feb 12 10:29:44 PST 2021


Hi eeglab list,

I recently started working with a 256 electrodes data set.
My concern today is related with the Artifact Subspace Reconstruction (ASR) for "bursty" activity. I have been trying to incorporate it to my pre-processing pipeline but it just doesn't work when I apply it on my full data set. I have noticed that it won't reconstruct the data unless I reduce the dimensionality of my dataset. If I remove lots of channels until I have 64 (for example) it would work as desired. I have tried this from the GUI and using the command line:

EEG = clean_rawdata(EEG, -1, -1, -1, -1, 10, -1);

Is there a minimum required number of data points needed to work with ASR?. Should I change some of the default parameters for this function?

I have rest eeg and task related eeg for each of my subjects, can I use the rest EEG to improve the performance of the method (once I get it running)?

Thanks a lot for your help and time

Sebastian Rueda



More information about the eeglablist mailing list