beamformer source localization for EEG

[Robert Oostenveld]

Hi Jess,

On 16 Jul 2009, at 19:57, Jess R. Kerlin wrote:
> I'm using a 128 electrode Biosemi EEG system, attempting to perform
> beam-former source analysis. I had go all the way back to Matlab 6.5
> and SPM 2, as well as some trial and error to get through it,  but I
> finally managed to get a nolte BEM  using mostly the same code as in
> the example script "BEM head model for EEG" script". However,
> prepare_leadfield throws the error "unsupported volume conductor
> model for EEG", which is clearly the case looking at the
> prepare_leadfield code.

The "nolte" model is a single shell realistic model for MEG. It is
based on the boundary of the inside of the skull, but is not a BEM
model and it is not for EEG. See http://fieldtrip.fcdonders.nl/references_to_implemented_methods
  and search for Nolte.

The boundary element method can be used for EEG volume conduction,
using the prepare_bemmodel function.

> A few questions on this topic:
>
> Does fieldtrip actually support beamforming source localization on
> BEM models for EEG?

Yes, see Schneider TR, Debener S, Oostenveld R, Engel AK, Enhanced EEG
gamma-band activity reflects multisensory semantic matching in visual-
to-auditory object priming. Neuroimage. 2008 Sep 1;42(3):1244-54

> Or, given the coarser spatial resolution of EEG, is it better to use
> a simpler spherical model?

The beamformer method is rather sensitive to volume conductor
misspecifications. So for a simple dipole fit a concentric spehrical
model might be enough, but for beamforming I would recommend to use a
BEM model.

> Either way, does anyone have successful sample code for beamformer
> localization EEG data, start to finish?

you can get a template BEM volume conduction model from ftp.fcdonders.nl/pub/fieldtrip/standard_BEM.zip
. The geometry is based on the canonical MNI template ("collin27") and
it is defined in MNI coordinates (consistent with SPM).


> And for the developers:
> I love your work.

thanks

> I notice the move from SPM 2 to SPM 8 is in the development cue.  Is
> there any release coming soon which will use SPM 8 to make this
> process smoother?

SPM2 is not supported any more in recent matlab versions, so we have a
compatibility problem one way or the other. Fieldtrip does not use
that many functions from spm, but some of them are not compatible
between spm2 and spm5/spm8. The idea is first to make a fieldtrip/
external/spm2 folder with a subset of the old spm2 functions in it
that is enough to keep it running. Those spm2 functions then have to
be fixed for compatibility with the most recent matlab (simple:
replace finite with "isfinite"). Subsequently, the difference between
those functions in their spm2 and spm8 incarnation has to be
determined and fieldtrip should be fixed to reflect these differences.

best regards,
Robert

----------------------------------
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.