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