[Eeglablist] Clustering ICs across sessions but within subjects

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Wed Aug 9 10:57:45 PDT 2017

Hello Ghislain, some recommendations below, best wishes.

1. the STUDY framework probably can't handle your full design, but it is
receiving a major update soonaccroding to eeglab developers.
Please closely review all current STUDY documentation online to determine
if STUDY can create the analysis your're interested in.
My bet is that you you can create data to represent each person's session,
and each person's condition, and
then you should attempt to build some STUDY versions that compare some or
all of your conditions.
Try something simple first, like just focusing on  just the witihn or just
the the between factors first.
LIMO EEG plugin for eeglab should also be helpful, check out the
documentation on that.

2. ICA solutions from different days (different recording sessions) often
give similar IC results, but overall it's not recommended to mix them, as
solutions can be quite different across different recording sessions (but
not always).

3. You could take all the data that you have for one person, from one
recording session, and then ICA all that data for that single subject. This
would at least give you a valid ICA decomposiiton for that subject, for
that session. Then you can use that ICA info to clean the data of
artifacts, and/or analyze the IC information.

4. You can compare the ICA results from the same subject from different
recrdogin sessions. There you should see basically the same ICs, if it is
the same person.

5. Note that some researchers prefer to do ICA across all available
conditions for a subject, and some do ICA for specific conditions
(especially if the conditions are very different)

6. Gift-EEG from the Calhoun group may also be able to help you find solid
ICA solutions across your study, subjects and conditions.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20170809/7c4688b8/attachment.html>

More information about the eeglablist mailing list