Hi Pedro,
I have zero expericence with EEG, so I don't know what would be best for
that application.
I don't need to find a transformation between surfaces having
corresponding landmarks, I need to obtain the corresponding landmarks.
But until you register surface A to surface B (or both A and B to C),
there is no way to establish correspondence between a landmark on
surface A and the corresponding location on surface B. Registration
builds that bridge.
Caret uses landmarks to drive registration, and there is no command line
method for identifying them yet. You must draw them for each
hemisphere. There are other surface-based registration methods that
require less/no user input, e.g. FreeSurfer:
http://surfer.nmr.mgh.harvard.edu
It might be possible to jump into the processing stream with already
generated surfaces, but my hunch is that you'll need to put your T1s
through the full pipeline. With 237 scans, I hope you have a lot of CPU
power.
Donna
On 06/20/2007 09:30 AM, Pedro A. Valdes Hernandez wrote:
Dear Donna: I'm interested in the surface commonly used for EEG
problems. I have 237 T1 MRIs and I'd like to have an automatic command
line script to obtain those mean WM-pial surface (if this is the
propper surface for my application) and I'd like to have corresponding
points between those surfaces. I don't need to find a transformation
between surfaces having corresponding landmarks, I need to obtain the
corresponding landmarks. Can you help me to understand what software
should I download, what should I run and do?
Any help will be very appreciated
Thank you in advance
Pedro
On 6/20/07, *Donna Dierker* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
Hi Pedro,
Yes -- Caret does surface-based registration, as described in the
registration tutorial:
http://brainmap.wustl.edu/caret/pdf/Caret_5.5_Tutorial_Segment.pdf
If these two surfaces are from the same subject, then the core 6
landmarks (the set typically recommended for inter-subject
registration)
will not achieve optimal registration; in that case, additional
landmarks (e.g., STS, POS -- as many as you can reliably identify
across
timepoints) are recommended.
For each surface, you'll need to use Surface: Geometry: Generate
inflated and ellipsoid from fiducial, to generate inputs the above
tutorial expects. You might also consider trying John's new
method for
registration without flattening:
http://brainmap.wustl.edu/pub/donna/PUBS/Reg_no_flat_and_new_border_drawing.pdf
login pub
password download
I haven't tried it myself yet. I'm not sure whether your Caret
version
will include all the features you need to try this.
Finally, we tend to use the midthickness (fiducial) surface,
rather than
the WM/GM surface. We generate a midthickness surface by averaging
coordinates of Freesurfer-generated [lr]h.white and [lr]h.pial
surfaces.
Donna
On 06/19/2007 04:48 PM, Pedro A. Valdes Hernandez wrote:
> Hi:
> I' completely new in this mailing list so maybe my question is a
> little bit foolish
>
> I have two triangulated WM/GM surfaces, I'd like to know if I
can find
> corresponding anatomical points between them using CARET/SureFit...
>
>
> Thank you in advance
> Pedro
>
------------------------------------------------------------------------
>
> _______________________________________________
> caret-users mailing list
> caret-users@brainvis.wustl.edu
<mailto:caret-users@brainvis.wustl.edu>
> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu <mailto:caret-users@brainvis.wustl.edu>
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
------------------------------------------------------------------------
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
