[Eeglablist] [eeglablist] aliasing in pop_spectopo():

Matt Craddock matt.craddock at uni-leipzig.de
Wed Oct 5 12:45:44 PDT 2011


On 04/10/2011 22:15, Bradley Voytek wrote:
> Anders:
>
> This is due to overflow errors due to the BioSemi DC offset. You have
> to remove the channel means *correctly* before running spectopo.
>
> See my same question and answer here:
> http://sccn.ucsd.edu/pipermail/eeglablist/2008/002229.html
>
> And this conversation:
> http://sccn.ucsd.edu/pipermail/eeglablist/2008/002148.html
>
> Good luck and let me know if you have any more questions!
>
> ::brad
>

Just to add that I've noticed this doesn't always fix the sawtoothing 
problem in the continuous data. In one dataset I had, sawtoothing was 
still visible after removing the channel means as above. In any case, 
once the data was epoched and baseline corrected, the sawtoothing 
completely disappeared. There was a big change in the DC offset on one 
channel at a particular point (I don't know why, but there was!), so 
subtracting the channel mean left half the data well below zero and half 
well above it - the big change in DC offset was no longer a problem 
after epoching, because each epoch was centred around a local mean 
rather than the global channel mean. (I was only looking at spectopo 
before epoching because I was wondering why it was so different before 
and after.) In general, the channel mean subtraction Brad suggests will 
work fine; but if it doesn't, something like the above might be why.

Cheers,
Matt



More information about the eeglablist mailing list