Hello, I've been looking through the archives and still have some questions about getting the best data for GM volume, WM volume, and the volumes of each sub-cortical structure. This is lengthy, but I'm trying to save a lot of future man-hours that I wasted on the first dataset I was doing.
*1)* I ran recon-all -autorecon-all overnight, not knowing that I could make edits in the middle of the process. Does this mean that, e.g if I need to add control points, I have to do / recon-all -autorecon2-cp -autorecon3, /and then make edits to the new wm.mgz and do /recon-all -autorecon2-wm -autorecon3/, *then* make edits to the pial surface and run everything to the end? Or, if I add control points, then only run recon-all -autorecon2-cp, this will be acceptable? And then edit wm.mgz, and run /recon-all -autorecon2-wm/? And so on? * 2)* This brings me to another head-scratcher for me: wm.mgz. Should all subcortical structures fall within the main surface (including ventral DC)? Sometimes the main surface cuts into the hippocampus/amygdala, so do I have to add/delete voxels so it goes around it? I would think the pial and main surfaces should exclude these two structures. The pial and main surfaces are ~identical around the midline and subcortical structures; is this the desired result? If i want to change the main surface in general, what is the significance of the white voxels vs. grey voxels? I think I understand that the white voxels are there just so the surface will go around it. So, in theory, if I wanted the main surface to include all of the hippocampus, I could fill in white voxels for the hippocampus, and then run recon-all? If I delete grey voxels, will this change the main surface (I suppose this would be like the opposite of control points)? I read in some places that the WM volume produced by mris_anatomical_stats is taken from all the voxels that are "on" in wm.mgz. If this is the case, then the ventricles are included? Some parts of the dura are "on" in wm.mgz (although do not fall within ?h.white); would I have to go slice-by-slice deleting these annoying voxels? * 3)* Is the data from mri_segstats and mri_anatomical_stats reliable? I've seen threads saying mris_wm_volume and mris_volume are better, since the GM and WM data are calculated using the surfaces. I don't have the former (I'm using FS v3.0.4). I noticed the WM volume given by /mris_volume rh.white - /{sum of volumes of subcortical structures} is very close to that given by mri_segstats, which is similar to using mris_wm_volume. Is this correct?
So, in the future, when I do recon-all for a subject, is it best to do it step-by-step, and make the edits as I go along?
Thanks for the help, Chris