Hi Donna,
Thanks for your reply. Acturally I'm a new user of CARET and I'm learning it
from the tutorial. I already fix the problem of the space and coordinates, and
now I understand I don't need to get flat maps for registration. Also I don't
want to use flat map because I tried two subjects, the flat map i got is really
... ugly...
But in this way, if I only have the fiducial surface, spherical surface, I need
to draw all the 6 landmarks manually, right?
With te 6 landmarks I draw, I can do the spherical registration. Is this a
right way to do it?
Thanks.
Jidan
>
> Message: 2
> Date: Thu, 17 Sep 2009 08:43:45 -0500
> From: Donna Dierker <[email protected]>
> Subject: Re: [caret-users] do spherical registration using freesurfer
> generated surfaces
> To: "Caret, SureFit, and SuMS software users"
> <[email protected]>
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset=GB2312
>
> Hi Jidan,
>
> See inline replies below.
>
> Donna
>
> On 09/17/2009 02:06 AM, z?? 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
> [email protected]
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>
>
> End of caret-users Digest, Vol 72, Issue 7
> ******************************************
_________________________________________________________________
Messenger安全保护中心，免费修复系统漏洞，保护Messenger安全！
http://im.live.cn/safe/_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users
