Hi Alessia
when you say "the image was rotated" do you mean you actually transformed the image, or you simply changed the image header to reflect a new orientation? I wouldn't think the latter would have a big effect, but the former will involve an additional image interpolation (blurring) and will definitely change things. Same question for 3. Did you include an extra interpolation?
cheers Bruce
On Thu, 5 Jun 2014, Alessia Giuliano wrote:
Dear FreeSurefer team, in order to verify the reliability of the volumes of Corpus Callosum (CC) subregions estimated by FreeSurfer I have applied the recon-all on the same subject but in three different situations:
- when the image was not preliminarly rotated;
- when an initial soft rotation was manually performed with SPM;
- when an initial rigid coregistration in MNI space was performed with SPM.
Although the differences in image orientation between 1, 2 and 3 before the implementation of FreeSurfer were really small, the differences in the volumes of the CC subregions between 1, 2 and 3 are notable.
How can I base on this evident variability my volumetric analysis of CC subregions?
Do you have any suggestions to improve my approach to the CC segmentation?
In order to make my results clear, I attach you a recapitulatory page and I send you a link to the FreeSurfer output in 1, 2 and 3 (https://www.dropbox.com/sh/k1z5o0e6sq90qq0/AABx38QxKvxDhUe2lV9imaa9a).
Thank you,
Alessia Giuliano