Nonlinear registration certainly generates a resultant image that looks more
like the template.  In my opinion, it results in better alignment of
structures and makes group analyses (e.g. fMRI, diffusion) more accurate,
however this opinion is not universally shared (yet).  High dimensional
nonlinear registration certainly has the potential to mismatch anatomical
structures that really are very different from eachother.  For example, lets
say that a subject has two gyri at a particular location in which the
template has only a single gyrus.  Nonlinear registration might fuse the two
gyri to make the image match the template, but this might obscure a real
biological difference.  That being said, FNIRT works very well in my
experience and I have not noted this sort of problem with it.  

 

Peace,

 

Matt.

 

  _____  

From: [email protected]
[mailto:[email protected]] On Behalf Of R. Matthew
Hutchison
Sent: Wednesday, May 12, 2010 6:52 PM
To: Caret, SureFit, and SuMS software users
Subject: Re: [caret-users] FSL and Caret

 

Thanks for the help everyone.

Wolf: I was wondering why you chose to do a nonlinear transform (FNIRT) as
opposed to the linear (FLIRT) transform.
Is non linear better when registering to an atlas?

Thank you all again, 

Matt Hutchison

On Wed, May 12, 2010 at 1:01 PM, wolf zinke <[email protected]>
wrote:

Hi Matthew,



On 05/12/2010 09:51 AM, R. Matthew Hutchison wrote: 

Hi group,

I am starting with Caret and have gone through most of the tutorials using
the data sets provided. The images are fantastic.
The part I seem to be missing is actually getting my activation maps so that
I can view them on the inflated surfaces.

If you use the menu 'Attributes', you'll find at the bottom 'Map Volumes(s)
to Surface'. I guess that is what you want. Took me a while to find it as
well ;-) .





I was wondering if anyone has a guide, steps, or hints in converting the
output files from FSL FEAT (registered to a standard atlas using FSL FLIRT)
to files viewable with Caret.
It seems this is not included in the tutorials.

T map macaque data of my individual monkeys to the caret F99 atlas, I used
FSL's FNIRT for a nonlinear transformation. Doing this once in a way, that I
was somehow satisfied with the result allows me to apply the transformation
on the statistical maps, if they are in the individual monkey space.
However, You might want to use the F99 atlas already as standard template in
the Feat analysis (if it yields convincing results).

I attached a small script that transforms a set of nii.gz-Nifti files found
within a directory into a new space according to a Flirt transformation
matrix file (linear) or a 3D volume with warp-coefficients (Fnirt,
nonlinear). Maybe this gives you an idea. I cannot guarantee that it is
bugfree, but at least It did work for me.






A search of the archive does not turn up any recent information on the issue
that I could find.

I am actually using macaque data but I figure the steps should be the same
as with the human but using the F99 atlas.
Perhaps I am just missing something obvious? 

I understand that my data must be registered to the atlas space that I wish
to use in Caret.
But which FSL file should be used and what is the next step so that I can
view them? 
If it is registered to the atlas then there should be no need to segment the
volumes again right?

You can use the zstat* images in the stats subdirecory of your copes as
statistical map that you transform to the standars F99 space. In this case
you can utilize the F99 segmentation, without segmenting the individual
brains. However, in my opinion you are loosing the specificity of the
localization, which in my view is one of the huge advantages of primate
fMRI. So I do both, mapping data on individual brains, but also map all
individuals on the standard atlas for comparision.






Thank you for any help (and patience for a novice),

I hope this gives you some pointers. Good luck anyway,
wolf





R.Matt Hutchison


-- 
R. Matthew Hutchison, PhD. Candidate
Centre for Functional and Metabolic Mapping
Robarts Research Institute
Cuddy Wing - 9.4T Suite
P.O. Box 5015, 100 Perth Drive
London, Ontario, Canada  N6A 5K8

 
_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users
  

 


_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users




-- 
R. Matthew Hutchison, PhD. Candidate
Centre for Functional and Metabolic Mapping
Robarts Research Institute
Cuddy Wing - 9.4T Suite
P.O. Box 5015, 100 Perth Drive
London, Ontario, Canada  N6A 5K8

_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users

Reply via email to