[Eeglablist] Pipeline of processing to optimize ICA for artrifact removal

[Modestino, Edward J *HS]

Dr. Makeig,

You mention the following formula below to determine how much data is needed to optimally use ICA.

#timepoints/(#channels)^2

So, if I have the following for each condition: (data was recorded at 1,000 Hz and I have about 186 one second epochs)

(1,000 time points *186 epochs)/(53 channels)^2

186,000/2,809 = 66.2157351

Below you mention it needs to be  much more than 1 or even as high as 30 for 256 channels.  What exactly is the cut off for optimal ICA using this formula? I want to be able to use this formula to determine for each dataset I have, ranging up to 128 channels and recorded at different sampling rates, what is enough data to run ICA optimally.

Thanks,
Dr. Modestino

Edward Justin Modestino, Ph.D.
Postdoctoral Research Associate
Ray Westphal Neuroimaging Laboratory
Division of Perceptual Studies
Department of Psychiatry and Neurobehavioral Sciences
University of Virginia
Email: ejm9f at virginia.edu<mailto:ejm9f at virginia.edu>

From: Scott Makeig [mailto:smakeig at gmail.com]<mailto:[mailto:smakeig at gmail.com]>
Sent: Thursday, February 23, 2012 11:34 PM
To: Arnaud Delorme
Cc: Modestino, Edward J *HS; eeglablist at sccn.ucsd.edu<mailto:eeglablist at sccn.ucsd.edu>; Kelly, Edward *HS; Ross Dunseath
Subject: Re: [Eeglablist] Pipeline of processing to optimize ICA for artrifact removal

Some additional >> comments,   Scott Makeig
On Thu, Feb 23, 2012 at 9:04 AM, Arnaud Delorme <arno at ucsd.edu<mailto:arno at ucsd.edu>> wrote:
Dear Mosdestino,

1. Difficulties with ICA. When removing ICA components, one of the main concern is the quality of your decomposition. We are currently working on tools to assess this quality although this can be tricky because of the large inter-subject variability. In the meantime, if you have multiple components for each type of artifacts, this is usually a sign that the quality of your decomposition is poor. One of the main factor to increase quality is to increase the amount of data and also high pass filtering if you have large offset in your data channels.

>> Modestino, you do not mention the data length he is using. An important value is #timepoints/(#channels)^2 ... this should be much more than 1 (even as high as 30 for 256 channels, in our (limited) experience -- I hope we will run a numerical experiment on this soon...  Running ICA on insufficient length data is the most common problem in applying ICA successfully.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20120301/8f44f9d7/attachment.html>