extract surface ROI center mass from .nii file


I am familiar with afni’s volume-based tools, but I have no shifted toward using surface-based tools from freesurfer and CONN toolbox. In order to upload an visual cortex atlas consisting of 50 ROIs I first converted it to a .nii file. I’d now like to determine the coordinates of each ROI center of mass. For a volume-based data set I would use something like 3dClust. However, SurfClust does not allow .nii files.

Is there a way to obtain surface ROI information from .nii files? Or to convert a .nii file to the spec and surface files required for SurfClust?


This is a little confusing to me. Are you trying to find a center for a surface, a patch on a surface or a volume? Surfaces are typically stored in GIFTI files ending in .gii. NIFTI files are most often used with volumetric data. Patches are usually a list of nodes in a niml.dset, 1D.dset, .1D and sometimes in .gii format as a “functional” GIFTI.

We had a poster on different ways to compute centers at last years OHBM.


If you provide some more detail, I’d be happy to point you in the right direction.

Thanks for your response Daniel. I, too, am confused. I am using an atlas of visual cortical regions (Wang et al 2015, Cerebral Cortex) where each regions is assigned one integer value. I am hoping to obtain the center mass for each visual cortical ROI. I had to use CONN toolbox, which required me to import an *.annot file to use the atlas. CONN toolbox then output a *.nii file. Since I’m more familiar with AFNI (volume data) I thought I could use this to obtain the ROI centers. But from your response it seems like I should start earlier in the process and work from a *.gii file.

Which AFNI/SUMA program would this *.gii atlas file be fed to so that I could extract the ROI center coordinates?


OK, that makes things a little clearer for me. As it turns out, I’ve spent quite a bit of time recently with that atlas. It is distributed with both surface and volumetric versions with probabilistic maps too in both formats. While it might not be obvious how to use the surface versions, I think that version is better in terms of contiguity and mapping to the MNI 2009c template or to each individual subject. PM me if you would like to chat about it, and we can set up a Zoom meeting to discuss.