[caret-users] do spherical registration using freesurfer generated surfaces

Donna Dierker donna at brainvis.wustl.edu
Fri Sep 18 09:26:07 CDT 2009


Traci,

I do have some scripts I could make available now, but in two-four
weeks, I'll have more general purpose ones. The most complete set were
actually developed for a data set segmented in Caret. But they include
the auto-landmarks, border variability (QA), generating scenes/captures
showing the borders, and running registration via command line. There
also are several versions of a "preborder" script that covers importing
from Freesurfer, but most of these used anatomical data that was in
711-2* space before feeding to Freesurfer. Your data is unlikely to fit
that category.

Next week, I'll adapt these import scripts to a dataset more like what
yours is likely to be. When I am done, I can post the resulting scripts.

Can you ping me in two weeks?

Donna

On 09/17/2009 04:07 PM, Sandoval, Traci I wrote:
> Donna, I would love to use these scripts that aren't publicized yet. How can I attain said scripts?!! Thank you for all your work and questions answered. Traci
>
>
> Traci Sandoval
> Research Assistant
> NeuroPsychometric Research Lab
> School of Behavioral and Brain Sciences - University of Texas at Dallas 
> Center for BrainHealth - University of Texas at Dallas
> Department of Psychiatry - University of Texas Southwestern Medical
> www.utdallas.edu/research/nprlab
> 972-883-3270
>
>
>
> -----Original Message-----
> From: caret-users-bounces at brainvis.wustl.edu [mailto:caret-users-bounces at brainvis.wustl.edu] On Behalf Of Donna Dierker
> Sent: Thursday, September 17, 2009 8:44 AM
> To: Caret, SureFit, and SuMS software users
> Subject: Re: [caret-users] do spherical registration using freesurfer generated surfaces
>
> Hi Jidan,
>
> See inline replies below.
>
> Donna
>
> On 09/17/2009 02:06 AM, z丹丹 wrote:
>   
>> Hi,
>>
>> I already have fiducial surfaces generated by Freesurfer, which are
>> all in the MNI space. I want to do spherical registration to an atlas
>> spherical surface. The steps i need to do are:
>>
>> 1, generate the ellipsoid, spherical surfaces, and flat maps.
>> 2, create the 6 landmarks on both individual and atlas surface.
>> 3, do spherical mapping.
>>
>> Are the procedure right? Thanks! 
>>     
> Funny you should ask. We (the Van Essen Lab) are using Freesurfer for no
> less than five projects, and we have shell scripts that streamline much
> of the import, border drawing, QA, and registration. There is a feature
> we call "auto-landmarks" that draws the borders reasonably well. Many
> borders need touch-ups, but a single rater fixed 72 hemispheres' borders
> in 1.5 days. If you have only a handful of subjects, it's not clear
> whether it would be worth it for you to use our methods. But if you have
> dozens, then they would save you a lot of time.
>
> We haven't broadly publicized these features/scripts, because they're
> still in the final stages of development and testing. But they have been
> working well for us.
>
> Even if you don't use auto landmarks or these scripts, one question
> stands out in your steps above: Do you need a flat map? No, not really.
> It certainly is easier to draw landmarks on the flat map than the
> sphere; however, I'm not sure it's worth the hassle. Then again, if
> you're not using the scripts, it might be the shortest path to registration.
>   
>> In fact before I move to the 2nd step, I was stuck in the flat map
>> generation. I met two problems:
>> 1, As my surface is in MNI space, the surface orientation is not right
>> in the CARET software. I found that the anterior and posterior are
>> reversed with superior and inferior. Is there any method to roate it
>> into the right orientation in CARET?
>>     
> Yes, but this is not consistent with my experience of MNI space nor
> Freesurfer surface orientation. While MNI volumes are sometimes stored
> using left-handed coordinate systems (what some call radiological
> orientation), the coordinates are always reported with -x being left, +x
> being right; -y being posterior, +y being anterior; -z being inferior,
> +z being superior. I've never seen a Freesurfer surface with the Y and Z
> axes flipped -- not even ones written in MNI space.
>
> It's easy to flip a surface from LAS to LPI
> (http://brainmap.wustl.edu/OLD/SureFit/orient.html) like this:
>
> caret_command -surface-apply-transformation-matrix $COORD $TOPO
> $FLIPPED_COORD -matrix "1 0 0 0 0 -1 0 0 0 0 -1 0 0 0 0 1"
>
> But I really would not brush this off. I would get to the bottom of why
> the supposedly MNI surface is not in the expected orientation. The
> evidence suggests it is not in MNI space, as asserted.
>   
>> 2, The other problem is , I found that the origin of my surface is not
>> AC. So when I flatten the surface, it can't find the correct medial
>> wall to cut. Do you think there is a way to solve this too?
>>     
> This is more evidence that the surface is not really in MNI space. How
> did it get transformed to MNI?
>
> Note that the MNI origin is not exactly the AC; see
> http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach for details. But
> it is certainly close enough to the AC that the compressed medial wall
> view should be centered over the medial wall. The fact that it is not
> indicates a problem with your transformation to MNI space.
>   
>> Thanks a lot!
>>
>> Jidan
>>     
>
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>   




More information about the caret-users mailing list