[FieldTrip] Projection from sensors space to segmented brain surface
johanna.zumer at donders.ru.nl
Wed Jun 20 18:22:48 CEST 2012
I would like to further clarify the options for this question. Perhaps
could you suggest which inverse algorithm you had in mind?
For example, it is possible to use beamforming limited only to voxels
defined as lying on the cortical sheet (i.e. restrict .inside to be only
the gray matter). This is possible, but you might miss out on the spatial
peak of the source lying just 5mm away to the inside a bit. Thus, it
would be better to do as Eelke suggested and apply beamforming to the whole
inside of the brain, and then project those values to the surface for
visualisation (which is easily done with the cfg.method='surface' option in
Alternatively, you can use a min-norm type of inverse method (e.g. sLORETA
or MNE) which are commonly used to reconstruct onto cortical surface voxels
only. This works because the min-norm algorithms force all activity to lie
in the region over which you select to reconstruct. Thus, a source 5mm
inside of the surface will automatically get projected to the surface.
I also agree with Eelke you should not just project sensors to surface
without doing a specific type of inverse method.
2012/6/20 Eelke Spaak <eelke.spaak at donders.ru.nl>
> Dear Andrea,
> If I understand your question correctly, you just want to *visualise*
> sensor space activity on a brain surface, without doing source
> reconstruction first? I would advise against this. Time-frequency
> power values of your planar gradient data reflect the spectral
> characteristics of the magnetic gradient at the location of your
> sensors. These sensors are not located at the surface of the brain,
> but quite some centimeters away from it. Therefore, visualising the
> data as if it were actually recorded on the brain surface, without
> doing a proper transformation, is potentially misleading. It is true
> that people tend to display all sorts of data in a 'topoplot' fashion,
> but in that case it is clear to the audience that you are looking at a
> representation in sensor-space. Have a look at ft_topoplotTFR for this
> approach, if you're not already familiar with it.
> If you want to display your time-frequency activity on a cortical
> surface (which is of course a perfectly valid thing to want to do),
> you first should map the sensor activity to source space. This mapping
> is non-trivial and can be done by e.g. beamforming. See
> http://fieldtrip.fcdonders.nl/tutorial/beamformer for a tutorial on
> this. That tutorial ends with a surface plot.
> Hope this helps!
> On 20 June 2012 17:26, andrea brovelli <andrea.brovelli at univ-amu.fr>
> > Dear Fieldtrip users and developpers,
> > I would like to project and visualise time-frequency power values
> > on planar gradient data at the sensors level onto the brain surface
> > segmentated from the anatomial MRI of the subject.
> > Can anyone suggest me how to do it in Fieldtrip ? Can anyone share the
> > scripts ?
> > Does anyone see any inconvenient to this visualisation ?
> > Thanks a lot for you help.
> > Andrea
> > --
> > NEW EMAIL: andrea.brovelli at univ-amu.fr
> > _______________________________________________
> > fieldtrip mailing list
> > fieldtrip at donders.ru.nl
> > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip