[Eeglablist] Edge artifacts with newtimef (problem with baseline)?

[Arnaud Delorme]

Dear Ana,

yes 'winsize' is the size of the moving window. Based on the size of the window and the number of wavelet cycles at the lowest frequency, the actual lowest frequency is computed. For example, for a window size of 1000 ms, if you have a 3 cycle window at the lowest frequency, the lowest frequency will be 3 Hz (3/1).
Alternatively, if you set the lowest frequency, the window size is going to be computed accordingly.

Arno

On 31 Mar 2013, at 17:06, Ana Navarro Cebrian wrote:

> Thank you very much Arno. I understand this better now. So I believe that if I set the baseline from -1500 to 0, for example, the analysis may be taking values greater than 0 (after the stimulus) to calculate the baseline. In this case, the analyzed baseline would go from -1500 to 148? Should I then use -412ms as the minimum possible value for the baseline?
> 
> Also, unrelated to that, what is the purpose of the 'winsize' parameter? Is it the same as the moving window, meaning it would overwrite the 824ms?
> Many thanks,
> Ana
> 
>  
> 
>  
> 
> 
> Subject: Re: [Eeglablist] Edge artifacts with newtimef (problem with baseline)?
> From: arno at ucsd.edu
> Date: Thu, 28 Mar 2013 10:20:42 +0100
> CC: eeglablist at sccn.ucsd.edu
> To: sabato45 at hotmail.com
> 
> Dear Ana,
> 
> at the lowest frequency, your moving window size is 824 ms, so if your epoch starts at -1500 ms, the center of the first window is located at -1088 ms. Because we only get spectral estimates at the center of windows, this is why you "loose" data on the edge. 
> 
> And this is what we generally mean by "border" effect (although it can also be what Makoto was talking about). I have not heard of edge artifacts. However, you must remember that the window you use at the lowest frequency include a total of 824 ms of data although it is tapered (multiplied) by a gaussian (in the case of Morlet wavelets) to avoid window edge effects.
> 
> Best,
> 
> Arno
> 
> On 21 Mar 2013, at 00:11, Ana Navarro Cebrian wrote:
> 
> Dear eeglab users,
> I'm using newtimef to calculate the ERSP. My epochs are -1500 to 2000 ms long, and I'm using the time from -1500 to 0 as a baseline. 
> I've been told that a baseline from -500 to -200 ms would be more optimal to avoid edge-artifacts, since my baseline starts at the same point that the beginning of my epoch (-1500ms). 
> 
> I don't observe any edge effects in my data (with the -1500 to 0 ms as a baseline), but I don't have other way to verify this and I'd like to make sure of it. Also,  I believe that this is related to the fact that my ERSP plots only show from around -1088ms to 1586ms (instead of -1500 to 2000), but I would like to understand this better.
> Could anybody explain why I shouldn't worry about edge artifacts when using newtimef to calculate the ERSPs?
> Many thanks,
> Ana 
>  
> _______________________________________________
> 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/20130406/03baea10/attachment.html>