The honest answer to most of these questions is "I don't know"; however, see inline comments below for any hints I think are worth sharing.
On Jul 15, 2012, at 7:56 AM, Colin Reveley wrote: > Before I begin something that's going to be tough, may I just confirm: > > I've got a structural and other (diffusion related) volumes, all in register, > at 250um. > > I need to make wm and pial surfaces in freesurfer, average them, and then map > the structural (myelin) and diffusion stuff using matt's technique. > > the trouble is that these volumes don't fit in 256x256x256. they are 330 long. > > so to make FS surfaces I need to downsample the structural from 250um to 300um > > there would then be a Flirt matrix that moved this back to the original voxel > dimensions and volume dimensions. > > my question is: once I've downsampled the volume and made the FS surfaces > (and thickness, curv metrics and ribbon volume) You'll probably have to multiply your thickness maps by the appropriate scalar. Curvature won't be affected, I don't think. Ditto for sulc, if you use that. > can I use the Flirt matrix to move the FS surfaces back into the original > voxel size, and also transform the thickness, curv metrics and ribbon volume > as being 250um not 300? > > I know I can rotate and scale the surfaces in space, so they work with caret. > and I can transform so that 1mm becomes 300um (just a matter of changing > units) . > > but can I then transform from 300um to 250um ? > > that's not the same as just calling 1mm some other unit. There is an actual > difference in the relation of node to voxel in changing from 300 to 250. You can always apply an affine transform to the surface (either in the GUI or command line), but you know that, so I guess I don't understand your question. > will the thickness and curv metrics, and the ribbon volume, upsampled from > 300 to 250 still work with the surface meshes made from 300 data, in the > myelin mapping system? > > the motivation here is to get as close to a surface made from 250um data as > possible, and to map the 250um data which is detailed. > > if it's not going to work, I may as well just down sample everything to 500um > and keep it there. This will be faster and easier, but less detailed. Maybe try it the easy way and see if the results look sane. You might learn something that makes you less inclined to do it the harder way (e.g., it's not going to work no matter which way you do it). If it works, but you'd like finer resolution, then sweat these issues. > > thanks, > > Colin _______________________________________________ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
