Dear experts,
I would like to ask for help in explaining an unexpected difference in volume in subcortical structures between orig.mgz space and rawavg.mgz space scans.
I used FreeSurfer (v5.1, Mac 10.7.5) on a group of subjects in order to extract subcortical segmentation volumes. I then changed the wmparc.mgz file to nifti format and used fsl tools to extract the hippocampus and other structures. I then used tkregister2 to produce the orig-to-rawavg matrix and applied this to each of the subcortical structure volumes individually to get these segmentations in the original native T1 space. On visual inspection this seemed to work fine, but when I compared the volumes of the orig.mgz space subcortical structures to the rawavg.mgz space structures there was a large difference (e.g. left hippocampus is in the range of 5000mm3 in the orig.mgz space and in the range of 7000mm3 in rawavg.mgz space). I wasn't expecting differences of this magnitude as I thought the rawavg-to-orig transformation only involved reorienting and reslicing?
I compared volumes using fsl's "fslstats image -V" command.
Can anybody please explain these differences and recommend a way to move forward? Ideally I would like to use the rawavg.mgz space subcortical structures for further analysis but I would like to know that I can trust them first!
Many thanks in advance!
Sean
Hi Sean, there are two possibilities. First, FS includes partial volume correction when estimating the size of hippocampus (and the other subcortical structures). Second, if you take it from a 1mm3 space to a space with larger voxels, then many voxels in the edge will be indeterminant, and that could cause the change. What is the size of the voxel in the original slice prescription? What did you use to convert it? doug
On 2/8/13 10:13 AM, Seán Froudist Walsh wrote:
Dear experts,
I would like to ask for help in explaining an unexpected difference in volume in subcortical structures between orig.mgz space and rawavg.mgz space scans.
I used FreeSurfer (v5.1, Mac 10.7.5) on a group of subjects in order to extract subcortical segmentation volumes. I then changed the wmparc.mgz file to nifti format and used fsl tools to extract the hippocampus and other structures. I then used tkregister2 to produce the orig-to-rawavg matrix and applied this to each of the subcortical structure volumes individually to get these segmentations in the original native T1 space. On visual inspection this seemed to work fine, but when I compared the volumes of the orig.mgz space subcortical structures to the rawavg.mgz space structures there was a large difference (e.g. left hippocampus is in the range of 5000mm3 in the orig.mgz space and in the range of 7000mm3 in rawavg.mgz space). I wasn't expecting differences of this magnitude as I thought the rawavg-to-orig transformation only involved reorienting and reslicing?
I compared volumes using fsl's "fslstats image -V" command.
Can anybody please explain these differences and recommend a way to move forward? Ideally I would like to use the rawavg.mgz space subcortical structures for further analysis but I would like to know that I can trust them first!
Many thanks in advance!
Sean
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas Greve -
Seán Froudist Walsh