[Eeglablist] ICA - Mimimum # of Data Points

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Fri Jan 18 11:20:36 PST 2019

HI Janelle, a few notes below. Best wishes.

*The formula is somewhat of an ideal for dense-EEG + ICA, and the
generallystory is that ICA has a big stomach (is hungry), and more time
samples will provide ICA more information, and thus more valid ICs.
*In brief, my recommendation is to go ahead and try doing ICA on a few of
your participant files, and see what artifactual ICs you get. Then I would
recommend doing the same on a longer recording, and comparing.
Then you have a little more evidence regarding your question. It cannot
hurt to take a little extra time to explore this issue.
*There are at least several dozen EEG articles in the published literature
that use quite short time periods (e.g., 1 to 10 minutes) and ICA. This is
major journals, and usually for resting data.
*In my personal experience with dense-EEG (128 and 256) channels, but with
relatively short time samples (e.g., 5 min) one can see that ICA is able to
at least show artifactual ICs
*You do have the option of sparsing down your channels to 128 or 64
channels, which would help you to get closer to this rule-of-thumb formula,
and shuld give you ICs that are more valid
*In general, one can also think about why not to record for 15 minutes or
30 minutes if one takes the time to setup a 256-channel recording.
*you may also consider PCA (not recommended by eeglab group), alternative
cleaning techniques (such as the ASR technique in one of the eeglab plugins
from C.Kothe), and... one could also
use a library of artifact IC maps, and use those maps to hunt for similar
patterns in your data (without necessarily doing ICA on your data)
*you may also benefit from googling eeglablist and your topic for previous
similar discussion, and also from searching on google scholar for published
articles that use ICA and short time periods (for published precedents that
you can mimic safely for publication)
*generally speaking, more time is better for ICA, but I am not aware of
formal methods articles that focus on this topic.

On Fri, Jan 18, 2019 at 11:28 AM Janelle Crouch <crouchjh at sunypoly.edu>

> Is it reliable to use ICA with fewer data points than called for in the
> *ch*2x30 *formula?  With 256 sensors, it would be difficult to collect
> such a large number of data points as our acquisition time is about 7
> minutes.
> If ICA results from a lower number of data points would not be reliable,
> can anyone suggest their preferred method of artifact detection and removal
> that does not use ICA?
> Thank you,
> Janelle Crouch
> SUNY Polytechnic Institute
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20190118/ed7c701b/attachment.html>

More information about the eeglablist mailing list