Hi Jidan, See inline replies below.
Donna > Hi , > > Thanks for the answer. > >> Given these objectives, it should be possible to do what you want. >> >> Let's call your starting template fiducial 'TF.coord' Your deformed >> template fiducial surface, DTF.coord, also has the same number of nodes, >> I gather - allowing you to create and display a node-based deformation >> vector. >> >> When you register DTF.coord to TF.coord, you will get a new surface that >> will be named 'Resampled_DTF.coord' if you select 'Resampled_' as the >> prefix on the Deformation tab when you run spherical deformation. If you >> create a deformation field between 'Resampled_DTF.coord' and 'TF.coord', >> it should reflect the node correspondences resulting from the >> registration process. > > I'm a little confused about the "resampled" surface. So from the above, > you mean I should do spherical registration between "DTF.sphere.coord" to > "TF.sphere.coord", then I can get a fiducial surface which is > resampled_DTF.coord. This surface, I think you mentioned previously, it is > just a resampling surface having the same number of vertices with the > TF.coord. I checked this surface, it has exactly the same pattern with the > DTF.coord. Do you mean the resampled_DTF.coord surface reflect the > registration? Then the result is really not satisfying as when I > superimpose the resampled_DTF.coord with DTF.coord, there is really little > difference which didn't change the pattern at all. This is correct, and if you recall the link I sent you earlier to the web page showing the sanity checking fiducial captures, you will note that each capture shows a different folding pattern, because each is a different subject/hemisphere. But yes -- these "resampled" fiducials do reflect the registration output, and besides all being on the same "standard mesh," the nodes are in correspondence with one another. Node i will be in the temporal lobe of all subjects, in roughly the same location, but because folding patterns vary so much (e.g., STS may be forked in subject A, but not in subject B), there is no guarantee that the anatomical correspondence holds precisely to a micro level. Recall that if you want to bring something (e.g., parcellation schemes) from TF to DTF, you can apply the inverse deformation to do that. Finally, I would point out that once you have resampled_DTF.coord on the same mesh as TF.coord, then you can use Surface: Measurements: Distance surface A to B to compute the dx, dy, and dz between the two. This is not something we often do, but I don't understand what you want. > > > > Do you think I do it right in the way above? > > > > >> Bear in mind that the deformation is unlikely to perform well unless the >> landmarks you choose are close to and effectively surround whatever >> local deformations were used to generate DTF.coord. Thus, if you >> created, say, a restricted deformation of one region (e.g., smoothing of >> the frontal pole), you would not obtain good registration using only the >> standard 'core 6' landmarks. >> > > > > My deformation filed is base on the global change, so the DTF.coord is not > restrictly deformed in one region. But from the resampled_DTF.coord, I > really didn't see much change. I think there should be some step wrong for > me. Hope for your suggestion. No, you didn't do anything wrong; the registration worked fine. This is expected behavior. I use these resampled fiducials, for example, to do coordinate-based morphometry. I want to retain the original subjects' folding patterns, but have all subjects' surfaces in register (i.e., node correspondence holds up across subjects) in order to do coordinate-based group analyses. > > >> Let me know if this is unclear. >> > >> David > > > > Thanks, > > > > Jidan > > _________________________________________________________________ > Á¢¿ÌÏÂÔØ×îаæ Windows Live Messenger£¡ > http://www.windowslive.cn_______________________________________________ > 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
