[FieldTrip] Please help me get component dipole fitting results using custom anatomy into normalized template for EEGLAB clustering.

Chadwick Boulay chadwick.boulay at gmail.com
Mon May 6 07:36:18 CEST 2013

Dear Fieldtrip users,

Using the subject's MRI and recorded EEG electrode locations, I've 
managed to segment the MRI, create the volume conduction model, align 
the electrodes, create a sourcemodel using a grid warped from the MNI 
template, create the leadfield, and perform ft_dipolefitting on IC 
weights (imported from EEGLAB). The geometry, electrode positions, and 
mri all line up (coordsys='ctf', unit='cm') and the calculated dipole 
positions are what I would expect.

[Aside: All of my topoplots have the electrodes positioned correctly 
relative to each other but the nose and ears are 90 degrees from where I 
would expect them. Do the topoplot functions not know how to deal with 
ctf coordsys?]

How can I normalize these dipole locations into the template brain so 
that I may perform component clustering in EEGLAB?
I have some ideas but I don't know how to implement them.

1) Calculate the transformation matrix to normalize the original MRI to 
the template MRI, then apply that transformation to the dipole positions 
and momentums. I'm guessing I need to use mri_norm = 
ft_volumenormalize(cfg, mri); but then I don't know how to use 
mri_norm.transform and mri_norm.initial. Further, the output of 
ft_dipolefitting doesn't seem to be suitable for ft_transform_geometry.

2) Since the sourcemodel grid is warped from the template brain, I can 
find the sourcemodel grid index closest to the dipole location then set 
the new dipole location to be the location of the template grid at the 
same index. I don't know how to do this for the momentum but I probably 
don't need the momentum for clustering.

3) Start over from the beginning, normalizing the MRI and the electrode 
locations as soon as possible then performing all subsequent 
calculations using the normalized geometry.

I searched the documentation and the list archives but I could not find 
an answer. I am surprised because I expected dipole normalisation to be 
a common requirement. Maybe I am missing something simple.
Thank you in advance for any help.

Chadwick Boulay
JSPS Postdoctoral Fellow
Keio University
Yokohama Japan
boulay at brain.bio.keio.ac.jp

More information about the fieldtrip mailing list