Hi Johanna,
On Fri, May 14, 2010 at 4:41 PM, Johanna Zumer <[log in to unmask]> wrote:
> Hi Gareth and Vladimir,
> I've got my first questions on SPM ... i've searched the pdf documentation
> and email archives but can't seem to find the answers.
> I normally load fiducials via the CTF .hdm file. under the 'prepare'
> option, -> Sensors -> load meg fiducials/headshape, I tried selecting the
> .hdm file, but that doesn't get recognised by ft_read_headshape (type
> ctf_hdm).
Why do you need to load fiducials via Prepare? For MEG they'll be
there from the beginning. That option exists to load a headshape
measured with polhemus that is not part of the CTF dataset.
> Is it possible to load them in head coordinates or subject-specific MRI
> coordinates? I know there is the read_ctf_hdm.m file within the external
> Fieldtrip code.. how does/can that get called by SPM?
Fiducials should be in the same coordinate system as the sensors and
they are as I mentioned.
> Also, I already have the 275 'local' sphere centres computed with CTF
> software, also stored in the default.hdm file. Can those be loaded, instead
> of SPM computing them from the SPM-generated head mesh?
This is something that hasn't been done till now. There are two
options for where these can be used. (1) If Fieldtrip can use those to
create a head model (vol) than this vol can be used in SPM but one
needs to make sure that everything else is consistent with it (i.e.
the mesh, the sensors the toMNI matrix etc.). (2) Perhaps more simply,
if you can use them via Nutmeg to compute leadfields, these leadfields
can be provided to SPM directly if you save them in the same kind of
file that SPM uses and replace the file that SPM generates with it.
> I could just try loading them from a text file but it seems SPM requires
> them to be in MNI coordinates. From Nutmeg, I already have the fiducials
> in MEG head coordinates, subject-specific MRI coordinates, and also the
> SPM-generated MNI transform, hence MNI coordinates. But... for most
> subjects, the nasion in particular resides in a space outside where the MNI
> transform exists, so I get [0 0 0] or NaN for the MNI position for each
> subject (using spm_sample_vol), thus I can apply the mri2mni Affine
> transform to get an approximation rather than the full nonlinear mapping.
> And just for comparison, I looked over 12 subjects I had for one study, and
> the average MNI coords I got for them, (placing the fiducials as I described
> to Gareth on tuesday, with lpa and rpa 1cm away from pa in the direction of
> nas, with nas at the inner point between nose and forehead), had a mean of:
> 0.1 82.4 -38.8 (nas)
> -81.0 -9.4 -51.1 (lpa)
> 80.7 -8.0 -52.1 (rpa)
> and standard error:
> 1.0 1.1 4.6
> 0.6 1.4 7.1
> 1.3 1.5 7.1
> whereas the CTF_FIL points in SPM are similar, but particularly different in
> the z-direction for lpa/rpa:
> 1 85 -41 (nas)
> -87 -11 -62 (FIL_CTF_L)
> 87 -11 -62 (FIL_CTF_R)
The method I used at the time to mark the points used as FIL_CTF...
was not something very rigorous so it might well be that your
coordinates are more precise. You should just make sure that you used
the same convention as used at the FIL and if so it might be necessary
to update the points.
> Finally, I tried using the 3D Source Recon GUI to load mri/fiducials: then
> if I load the subject-specific MNI coordinates to the 'template', there
> seems to be an affine transform generated D.other.inv{1}.datareg.toMNI which
> I can apply to D.other.inv{1}.datareg.fid_mri_fid.pnt to get back to the MNI
> coords I loaded. However, if I load the same MNI coordinates to the
> subject-specific 'MRI', then that transform doesn't seem to bring me back to
> the MNI coords I originally loaded. Any suggestions?
See
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=SPM;6b20e904.1004
Vladimir
> Thank you for any suggestions, and I'm sorry if i'm asking questions already
> answered elsewhere.
> Cheers,
> Johanna
