Stability of ft_dipolefitting

[Paul Czienskowski]

Dear all,

as a pretest for my diploma thesis, which deals with the source
localization in EEG, I tried to generate some simulated EEG Data and fit
them both with the BEM model I generated them with and a BEM model I
generated from the icbm152 brain. The set of electrodes was generated
from the  10-20-system's instruction. I found that the  'real' brain
model fitted the data quite good, but the fitting with the other model
failed completely for it 'found' the dipole position on the way other
hemisphere pointing inwards instead of in the vertex direction. Since
this seemed quite  strange to us we decided to fit the data with one of
our other real brain models. Even if the dipole was located in the right
hemisphere now the fit was quite poor for it located the dipole in the
brain stem or in the medulla oblongata rather than in the auditory
cortex where it was simulated. When I didn't let the fit perform the
scan on the grid but let it run on some initial position in the auditory
cortex the fit with one of the other models performed a bit better but
still with a too small moment and still ~5 cm away from our position.
I know I won't be able to perform a perfect fit with another model but
it should be better than this - at least I thought - for I think we'll
never have a perfect model of a head and thus we wouldn't be able to
perform any reasonable fit if it was that unstable. Has anyone any clue
which could have been my fault. I know it's not easy without the data
and even the figures but maybe there's anything you already see from my
approach which was completely wrong. Some basic data: As I told you
before I used a 10-20-electrode-system, my models had 4 layers with each
about  1000 vertices ([999,1004] to be accurate), generated from a spm8
segmentation with a script inspired by
http://fieldtrip.fcdonders.nl/example/create_bem_headmodel_for_eeg . I
generated the models with OpenMEEG. For the grid search I used the gray
matter from the spm8 segmentation and down sampled it to a 5 mm grid.

Thanks in advance,
Paul Czienskowski

--
Paul Czienskowski
Max Planck institute for human development
Lentzeallee 94
14195 Berlin

Björnsonstr. 25
12163 Berlin

Tel.: (+49)(0)30/221609359
Handy: (+49)(0)1788378772

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