dics method for coherence analysis

Mon May 7 11:38:21 CEST 2007

Dear Feng,

You can save yourself some time by running sourceanalysis twice.

The first time you should specify cfg.keepfilter = 'yes' (and keep
cfg.randomization = 'no').

This will yield a source-structure containing a field called 'filter'. These
filters are computed using the cross-spectral density averaged across all
replications.

Subsequently you should plug these filters into the dipole grid by
specifying cfg.grid.filter = source.filter (could also be source.avg.filter,
I'm not sure, but it's easily checked by running the previous step). Then
you should additionally specify cfg.rawtrial = 'yes'.

If I'm not mistaken, when you then run sourceanalysis, this yields a
source-structure containing a field called 'trial', which contains for each
replication the source-reconstructed data, so it has dimensionality 1x the
number of trials. Each trial itself will be a structure containing the
fields csd (which in your case will contain the cross spectral density
between the dipole at that particular location, and the reference), and the
field pow, which contains the power at that location (not the reference).

In subsequent analysis steps you can test whether coherence is different
across conditions. This can be done parametrically, or non-parametrically.

Either way, I'd advise you to collaps your data across the two conditions
prior to computing your filters (so prior to calling sourceanalysis for the
first time). This yields so-called 'common filters', which as I said before
typically yield more robust results. The most convenient point to achieve
this, is already concatenating the data of your two conditions prior to
frequency analysis (so it should be done after preprocessing), by calling
appenddata: dataall = appenddata([], data1, data2).

You could then proceed to compute the filters, using the concatenated data.
Afterwards, you could call sourceanalysis twice using the frequency-data for
each of the conditions separately, yielding source structures for each of
the conditions. These can then be used as an input to sourcedescriptives
(should also be called twice). This will yield two source structures
containing an estimate of the coherence, and an estimate of the sem of the
coherence. This can then be tested parametrically with a t-test using
sourcestatistics. (It would be more correct if you correct for the bias in
the coherence estimate by applying some normalizing transform to the
coherence, see Bokil et al. J. Neurosci Methods 159(2): 337-345.)

You could also test non-parametrically. For this, I'd advise you to first
have a thorough look at the statistics-tutorial on the fieldtrip-website.
Moreover, you could consult Maris et al. J. Neurosci Methods 163(1): 161-175
for a way of testing coherence differences non-parametrically. However,
doing this on the source-level is not yet supported by fieldtrip in a
standard way.

Hope this already points you in the good direction.

Yours,

Jan-Mathijs

  _____

From: FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL] On Behalf
Of Rong, Feng (NIH/NIDCD) [V]
Sent: Monday, April 30, 2007 6:47 PM
To: FIELDTRIP at NIC.SURFNET.NL
Subject: Re: [FIELDTRIP] dics method for coherence analysis

Jan-Mathijs,

Thanks for the comprehensive feedback. It is really helpful. I have
downloaded the new version and will try.

As you have predicted, this procedure is time-consuming. May I know what can
I do to revise the script so that I can save some time?

Best,

Feng

_______________________________________________ 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.
http://listserv.surfnet.nl/archives/fieldtrip.html
http://www.ru.nl/fcdonders/fieldtrip/

----------------------------------

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.

http://listserv.surfnet.nl/archives/fieldtrip.html

http://www.ru.nl/fcdonders/fieldtrip/

----------------------------------

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.

http://listserv.surfnet.nl/archives/fieldtrip.html

http://www.ru.nl/fcdonders/fieldtrip/

----------------------------------
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.
  http://listserv.surfnet.nl/archives/fieldtrip.html
  http://www.ru.nl/fcdonders/fieldtrip/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20070507/43c95947/attachment.html>