[Eeglablist] clean_rawdata and ICA

Steven Dong (HARMAN Connected Services) v-jendo at microsoft.com
Mon Feb 4 13:27:41 PST 2019

Hi all

I am new to EEG and have a few questions about preprocessing and whether the Artifact Subspace Reconstruction (clean_rawdata) can be used for ERP or event-base experiment.

1.       For ERP experiment such as the oddball paradigm study could I use ASR to clean the data?

I tried it once on the continuous data and found that lots of markers were missing which means some of the data were removed, so I assumed using ASR will remove some data which make it not time-lock anymore. If not, what will be the best way to do auto channel rejection and removal with the script?

2.       If I am running an experiment which I am monitoring the mental workload across time, should I first slice the data into different conditions (i.e., for 3 mins is condition A, 3-6 mins is condition B, etc.) and do preprocessing and ICA individually? Or should I first do all the preprocessing and then slice the data? I think if I am using ASR I will have to slice the data first since if it remove some data the original timestamp might be shifted.

3.       In other word, I am wondering is there any difference if I treat each sliced data as independent data and run preprocessing/ICA individually versus run ICA on unsliced (all condition combined) data?

4.       Additionally, if I collect multiple data (different condition) from the same participant during one experiment session (cap is not removed), should the data be concatenated before running ICA? Could ASR still be used for this case?

Thanks for all your time reading through my questions.

Steven Dong

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20190204/20aa545f/attachment.html>

More information about the eeglablist mailing list