On 07/08/2016 02:21 PM, Arkadiy Maksimovskiy wrote:
Hello,
I apologize if the questions are trivial- I have checked the archives, but could not find anything on this.
I have run a t-test between two groups on cortical thickness, and used mri_glmfit-sim for pos, neg, and abs versions.
- If my FSGD is organized as:
Group 1 Group 2
and my matrix file is 1 -1
Would a blue result in an absolute version indicate that Group 1 < Group 2?
Blue would mean that Group 1 < Group 2
- When running the command: "less" on
"cache.th13.pos.sig.cluster.mgh", I am getting 1 significant cluster, however, when running the same command on the "absolute" version of this file, there is nothing significant. Do you know why this might be the case?
A signed test (eg, pos or neg) has a different voxel-wise threshold than an unsigned test (eg, abs). By default, FS computes statistics in an unsigned manner, so when you specify a cluster-forming threshold with abs, that is the actual threshold that is used. When you use pos or neg, the cluster-forming threshold is made more liberal to account for the fact that you are doing a signed test. For a cluster that is all negative, using neg will result in a more liberal threshold and so a larger cluster which may be more significant.
Much appreciated,
Arkadiy
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer