baseline, combine_planar, and freqanalysis

[Michael Wibral]

Dear Sangita,

the problem you raised is actually a tough one, that does not have a fully satifactory solution. The fundamental reason is the loss of spatial direction information in combineplanar that either affects interpretability of the resulting quantity or implies certain assumptions. Below I try to give you a quick explanation - the mathematical details are a bit lengthy.

Baseline correcting before combineplanar would either amount to the same as subtracting the baseline timecourse from the task time course directly - if done on the 4D (!!) object composed of the two complex fourier coeffients in the x,y gradient vector. Else, if you subtract amplitudes or power values per spatial direction separately a relatively uninterpretable quantity (because spatial angles between task and baseline would play a major role) results after combineplanar. This is because of the effect of different gradient field angles between task and baseline. So the only viable way seems to be to correct after combineplanar. The implicit assumptions then is however, that source configuration (orientation) does not change between task and baseline, this should be justified/discussed for the case at hand. This problem does not occur for axial quantities as there is no loss of spatial information there.

Hope this helps, please also let me know if you (or anyone else) disagrees!
Michael

PS: If anyone knows how to quantify spectral power of *arbitrary time varying vector fields* (not linearly or circularily polarized ones) uniquely and without assumptions on correlations between vector components please post it to the list.


---------------------
Von: "Sangita Dandekar" <sangita.dandekar at GMAIL.COM>
Gesendet: Oct 5, 2010 1:56:32 PM
An: FIELDTRIP at NIC.SURFNET.NL
Betreff: [FIELDTRIP] baseline, combine_planar, and freqanalysis

Thought I'd try reposting once, since I happened to send this on a weekend and didn't receive any reply. Sorry for the repost.
Thanks,
Sangi


---------- Forwarded message ----------
 From: Sangita Dandekar <[sangita.dandekar at gmail.com]>
Date: Fri, Oct 1, 2010 at 12:00 PM
Subject: baseline, combine_planar, and freqanalysis
 To: FieldTrip discussion list <[FIELDTRIP at nic.surfnet.nl]>


Hi,

I'm wondering about the optimal order in which to apply combine_planar and baselining (relative to a pretrial time window)  to time-frequency power spectra data.

Does it make more sense to do:

 A:
1. meg_planar
2. freqanalysis
3. combine_planar
4. freqbaseline (on a trial by trial basis?  see below)

or B:
1. meg_planar
 2. freqanalysis
 3. freqbaseline(on a trial by trial basis? see below)
 4. combine_planar


Also, are there any opinions on whether baselining on a trial by trial basis is preferable or  if it might be better to instead baseline with respect to the average of all pretrial baseline power spectra?

Thanks in advance,
Sangi






--------------------------------------------------------------------------- You are receiving this message because you are subscribed to
 the FieldTrip list. The aim of this list is to facilitate the discussion
 between  users of the FieldTrip  toolbox, to share experiences
 and to discuss  new ideas for MEG and EEG analysis.
 See also http://listserv.surfnet.nl/archives/fieldtrip.html
 and http://www.ru.nl/neuroimaging/fieldtrip.
 ---------------------------------------------------------------------------

---------------------------------------------------------------------------
You are receiving this message because you are subscribed to
the  FieldTrip list. The aim of this list is to facilitate the discussion
between  users of the FieldTrip  toolbox, to share experiences
and to discuss  new ideas for MEG and EEG analysis.
See also http://listserv.surfnet.nl/archives/fieldtrip.html
and http://www.ru.nl/neuroimaging/fieldtrip.
---------------------------------------------------------------------------
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Michael Wibral.vcf
Type: text/x-vcard
Size: 628 bytes
Desc: not available
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20101007/0207fb3e/attachment.vcf>