[FieldTrip] Dipole fitting

[Russell G Port]

Hi All,

I am currently struggling to fit dipoles to my auditory data. In short I
want to 'turn off' half the channels (so I only have one hemisphere)  (as
can easily be done in CTF tools DipoleFit). I thought this would be easy by
having the timelocked data that went into ft_dipolefitting, have
cfg.channel='MR' or 'ML'.
The issue I have is when I dipole fit this data, the dipole ends up in the
center of the head. I definitely have only left or right channels selected
(as checked in the output of ft_timelockanalysis). I thought of two ways
around this. A) I use an initial guess away from the center of the head
(with cfg.nonlinear ='yes') B) I have a source model with the medial -
lateral space starting starting away from the center (e.g. in CTF
coordinate Y = 2:1:7) [again with cfg.linear='yes']. While these approaches
normally work, I have a particularly stubbon couple of datasets, that still
think the the dipole is in the center of the head. When I check the
topoplot of the date being fit, it appears that the 'turned off/not
analyzed' side (the side not specified in ft_timelock analysis's cfg) is
completely red on the topoplot, which would not be a problem but there is a
small sink close to it, which is throwing off the dipole placement. Next I
repeated my dipole fitting using my altered source model with
cfg.nonlinear='no', but again this source and sink caused by the
contralateral hemisphere is pulling the dipole to the edge of my source
model space. Does anyone have any suggestions. When I try to dipole fitting
this data in CTF DipoleFit I do not have this issue (presumably because the
contralateral hemisphere is no contributing to the dipole fit).

Thanks
Russ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140322/39a20e7a/attachment.html>