Dear FreeSurfer gurus,
We are trying to conduct a simple group analysis, employing freesurfer. However, our aim is to use the software only up to the point where the functional data is projected to a common space, and from that point on export the data to a text file and conduct the analysis using an external mathematical package (R, http://www.r-project.org). We have received very useful information from this message board in the last 2 weeks assisting us in this endeavor, but are still stumped by part of the workflow.
At this point, we have succeeded in projecting our functional files (AFNI format) to the FreeSurfer surface representation so that each individual functional data is project to the individual registered surface. Given that each individual surface has a different number of surface nodes, we reasoned that to conduct a group analysis, it would be necessary to project the functional data not to the individual surface of each participant, but instead, to the registered icosahedron representation. The documentation of mri_vol2surf states that using the "icoorder" option, will result in mapping to an icosahedron with *prespecified* sizes. Specifically, we have tried projecting to an ico with order 7 (documentation states 163842 nodes in this reprsentation). However, when using this option and outputting to a "paint" (.w) file, we do not get this number of nodes for the icosahedron. Instead, different subjects' icosahedrons have different numbers of nodes. This is evident when the paint file is converted to and ".asc" file via mris_convert. Before we continue, we would like to know if this is a bug in mri_vol2surf, in mris_convert, or perhaps not a bug at all (i.e., different icosahedrons may indeed have different number of nodes by design).
Now that we find ourselves with no way to equate the number of nodes in each participants' surface representation we are faced with a quandary: How do we conduct the group analysis? It has been suggested to us that we use the surf2surf utility. The documentation of that utility mentions that both the src and target surfaces may be icosahedrons. If we do indeed need to use this utility, should we map data between icosahedrons, or would it be better to cut down one stage of interpolation and just directly project the surfaces of all subjects to one reference subject (e.g., subject b -> subject a, subject c ->subject a, and so on).
Thanks for any assistance, Emily