Hi Jidan
Lilla Zollei (ccd) has worked on some tools that we just published in IEEE TMI for applying the surface warp to an entire volume. I guess you could just warp the surface through forward and inverse transforms, but the transformed source will be guaranteed to lie within the target surface by construction, so I'm not sure how interesting it will be.
cheers, Bruce
On Fri, 24 Jul 2009, Michael Harms wrote:
Hi Jidan, It sounds to me that what you're asking is whether it is possible to invert the "mapping" from {lh,rh}.white to {lh,rh}.sphere, but apply that inversion to {lh,rh}.sphere.reg, thus generating a putative "{lh,rh}.white.reg" surface. I'm not sure that operation is possible, or if the resulting surface would even be meaningful.
cheers, -Mike H.
On Fri, 2009-07-24 at 10:20 +0800, Zhong Jidan wrote:
Hi Bruce,
For my 2nd point, I mean, after we get the lh.sphere.reg as the deformed subject, it should have similar sulco-gyral pattern as the template. I want to compute the distance between the deformed subject and the template, to see how close they are after the registration. In other words, I want to check the distance error. But the distance based on the sphere doesn't include the geometry information. I want to get the real distance in their white surface, which is between lh.reg.white and template.white. Also if I have lh.reg.white, it would be interesting to compute the real distance between the subject.white and lh.reg.white to see how far it moves.
But from the method you told me, "You can use the mapping to paint the geometry of one subject onto another folded surface if you want to visualize the geometric mapping (e.g. the curv of one subject on the white surface of another)." If "paint" means set a value for the vertex for visulization, (just like paint the functional information on a surface ) this would not make me get the real geometry of the .reg.white, right? When you paint the curvature information of the lh.sphere.reg and template.sphere to another white surface, you can only see the curvature difference on the corresponding point but can not see the real geometry of the two. I'm not sure whether I understand your surggestion correctly...This what I'm confused.
Thanks, Jidan
On Thu, Jul 23, 2009 at 8:12 PM, Bruce Fischl fischl@nmr.mgh.harvard.edu wrote: Hi Jidan,
we construct fsaverage by mapping the icosahedron vertices to a set of subjects, then assigning that vertex to the position that is the average of the tal coords of those vertices. Sorry, I don't understand the 2nd point. Maybe you can explain what you are trying to accomplish? cheers, Bruce On Thu, 23 Jul 2009, Zhong Jidan wrote: Hi Bruce, Thanks for your reply. I'm not clear about two points you mentioned. ->To represent a folded surface we either (1) map to an individual subject, or (2) use a volume transform such as tal and the vertex correspondence from the sphere.reg to find the average coordinate of a vertex. This is how we build the fsaverage and average7 surfaces. For the (2) point, what do you exactly mean? I?m confused. I thought your fsaverage is derived by averaging 40 subjects' volume image and then reconstruct the surface based on that. How can I use a volume transform such as tal and the vertex correspondence from the sphere.reg to find the average coordinate of a vertex? I'm totally lost with this sentence.. ->You can use the mapping to paint the geometry of one subject onto another folded surface if you want to visualize the geometric mapping (e.g. the curv of one subject on the white surface of another). For this one, I think you mean that, I can use one white surface as a base. Then paint the curvature information on the white surface to see the geometry. But this is not I want... To be specific, if I have a lh.sphere, lh.sphere.reg, they share the curvature information, and the only difference is the spherical coordinates. Then it is non sense to paint the curvature information to another white surface to see the geometry because there is only one .curv file. Another problem is, this lh.sphere may have different number of points with the white surface. ... Not sure whether I understand it correctly, hope for your suggestion. Thanks, Jidan On Tue, Jul 21, 2009 at 8:26 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu>wrote: Hi Jidan, the atlas only exists in spherical coords. To represent a folded surface we either (1) map to an individual subject, or (2) use a volume transform such as tal and the vertex correspondence from the sphere.reg to find the average coordinate of a vertex. This is how we build the fsaverage and average7 surfaces (we know this is a hack, but it's easy and a good visualization tool). You can use the mapping to paint the geometry of one subject onto another folded surface if you want to visualize the geometric mapping (e.g. the curv of one subject on the white surface of another). cheers, Bruce On Tue, 21 Jul 2009, Zhong Jidan wrote: Hi Freesurfer experts, I asked this question previously, but I found it problematic when displayed in your mailist. I'm sorry that the question still not solved and I feel sorry to trouble you again. In your sphere registration in freesurfer, the procedure is like: creating the template.tif by mris_make_template. The template you use in Freesurferis created by iterative registration of 40 subjects, according to "High-resolution inter-subject averaging and a coordinate system for the cortical surface, Fischl, B., Sereno, M.I., Tootell, R.B.H., and Dale, A.M., (1999). Human Brain Mapping, 8:272-284 (1999)". So, after the template generation process, you will get a .tif file which include the necessary infomation (like the means and variances of curv, sul from the aligned spheres). But,do you have the other information of this final template, such as the sphere representation, folded surface representation of this template? I know that under */subjects/fsaverage/surf, there are some surface representations of the average of the 40 subjects, but to my knowledge, they are just used for visulazation and are not the surface representation of the template.tif you used, am I right? 2, subjects' sphere registration to the template sphere In this process, we can get the deformed subjects spheres( *.reg ), which have a one-to-one correspondance to the original subject surfaces. Except the .reg sphere with the cuvature information, do you have any other form of representation of the deformed sphere? You know that there are other kinds of surface mapping methods, like Miller's Large Defformation deffeomrphic surface mapping, they just do surface mapping using the folded surfaces. After surface mapping, they will get the deformed folded surface which would be aligned with the template folded surface. With the deformed subject and template folded surfaces, they can tell directly which sulcus or gyrus is aligned well. So, for your mapping, when I get the deformed sphere, do you have any command or method to put the sphere back to the folded surface so I can see the suci and gyri directly? If you also have the surface representation of the template, then i can superimpose them to see how good the alignment is. If you think I didn't state this problem clearly, please refer to an example in the following: I found one reference using your sphere registration method. "Simpliÿÿed Intersubject Averaging on the Cortical Surface Using SUMA"Brenna D. Argall, Ziad S. Saad,and Michael S. Beauchamp"Human Brain Mapping 27:14 ÿÿ27(2006)" You may see the attachment in : https://mail.nmr.mgh.harvard.edu/pipermail//freesurfer/2009-May/010558.html In "Spherical Morphing" section, They mentioned that " Using the mris_register [Fischl et al., 1999b] routine, each individual subjectÿÿs surface was registered to the FreeSurfer average7 template prior to node number standardization. Standardization and averaging were then performed on the surfaces as described above" (using SUMA FYI). ---- From this part, I assume that all the deformed surfaces are in spherical representation. Then in the result part, in section "Intersubject Averaging of Functional Data: Different Surface Methods", they mentioned they " in order to compare the ACÿÿPC method to these more complex algorithms, the FreeSurfer program mris_register [used in Fischl et al., 1999b] was used to morph the cortical surface models to a predeÿÿned template, and these morphed surface models were then used to create a morphed surface average." In Fig7C ÿÿAverage surface created by averaging the same 28 subjects using mris_register standardization. You can see that they show the average surface in a folded surface representation, not a sphere. Could you give me a hint that how they do this since you only have a sphere representation of the aligned surface?-- Regards,
Jidan _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer