Hi Jingyue
you can use mris_make_template to create a folding atlas from your subjects (assuming you have surfaces, do you)? Once you have the atlas you can create downsampled versions using the
make_average_surface --ico <N>
script using N<7 (N=7 is our default 160K or so icosahedron. Every increment below 7 decreases the # of vertices by about a factor of 4).
cheers Bruce
On Wed, 29 May 2019, Jingyue Zhang wrote:
External Email - Use Caution
Dear Bruce:
Sorry I forget to tell you I'm using the gifti format 40 subjects dHCP data, so these are neonatal data, I guess existing atlas like fsaverage may not work well. So I want to create my own refere nce template and project the fMRI data on it. And I guess there should be 40 time series on each ver tex on the template corresponding to each subjects', and they come from mri_register or mri_surf2sur f? However, After that mail I found an atlas by (Bozek et al. 2018, https://brain-development.org/br ain-atlases/atlases-from-the-dhcp-project/cortical-surface-atlas-bozek/). The problem is about two m ain points: (1). Project my fMRI data from individual subject to template. (2). The vertex number is large (~6k-10k), so I want to downsample the atlas and then do the register but I don't clearly kno w how to do.
(1). If I focus on the existing atlas by Bozek et al, Can I use mris_convert to convert them into Freesurfer format and downsample it to ~10% of the vertex and then register each of my subject's time series onto it? If yes, is mris_decimate and then mri_surf2surf should be the right command?
(2). If I still want to create my own template. After generated lh.wholebrain.tif, and I get lh.sphere.reg for each subject. Then I think I could use mri_surf2surf to map the individual fMRI to t he template? Am I doing the right thing? And when and how should I do the downsample?
Really sorry for not clearly described my question in last mail! Still looking forward to hearyour reply, thank you!
Best wishes, Jingyue Zhang
Bruce Fischl fischl@nmr.mgh.harvard.edu 于2019年5月28日周二 下午10:52写道： Hi Jingyue
can you explain what your goal is in a bit more detail? My guess is you can use the existing atlas/template and not create a new one. As for the gifti conversion, I think mris_convert can handle it cheers Bruce On Tue, 28 May 2019, Jingyue Zhang wrote: > > External Email - Use Caution > > Dear Freesurfer experts: Hello! I'm a new student to surface based > fMRI analysis. Now I'm dealing with a functional parcellation problem. The > subjects in my dataset has different vertex number and I need to do a group > level parcellation, so I decided to make a group reference template. > My step is almost following thispost: https://www.mail-archive.com/freesurfer@surfer.nmr.mgh.harvard.edu/ms > g00420.html ,the lh.wholebrain.tif and lh.sphere.reg for two test subjects > have successfully generated. But I'm still wondering about these files and > what to do next to get my individual fMRI time series on vertex registerd to > the template. > The second question is, the vertex number (~100000) is much more > than my voxel based masked data (~50000). Neither is easy to do the > susequent analysis efficiently. So how should I correctly downsample it? > Should I downsample the template or downsample my surface before making > template? Will mris_decimate work? > The last question is my subject has gifti format sulc and curv data > (.shape.gii). I have tried to use MRI_read and MRI_write to write to an .mgh > file and add -sulc sulc.mgh for each command and they seems worked. Can I > convert them to file like .sulc? Or just keep the .mgh and -sulc sulc.mgh is > OK? > I'm looking forward to hear your reply! Really thanks for your help! > > Best wishes, > Jingyue Zhang > >_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer