Dear all,
I would like to ask you whether there should be an exact correspondence between vertex-wise thickness and mean thickness of clusters.
I'll explain you my problem. I did an analysis with 2 groups and 2 covariates. Then i run the mri_glmfit script and found clusters of greater thickness in controls compared to patients. When i extracted the mean thickness of these clusters and did a two sample t test i observed that the controls had greater thickness than patients, however not statistically significant, even after including in this analysis the 2 covariates.
Is this normal? How does freesurfer uses the covariates when performing an analysis?
Many thanks
It might be, it depends on how you defined the sign of the voxel-wise threshold. If you used "abs", then it is certainly possible.
doyug
Joana Braga Pereira wrote:
Dear all,
I would like to ask you whether there should be an exact correspondence between vertex-wise thickness and mean thickness of clusters.
I'll explain you my problem. I did an analysis with 2 groups and 2 covariates. Then i run the mri_glmfit script and found clusters of greater thickness in controls compared to patients. When i extracted the mean thickness of these clusters and did a two sample t test i observed that the controls had greater thickness than patients, however not statistically significant, even after including in this analysis the 2 covariates.
Is this normal? How does freesurfer uses the covariates when performing an analysis?
Many thanks
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Dear Doug,
Thanks for the reply.
Would you recommend using pos or neg sign, instead of abs for a group comparison?
Are my vertex-wise results showing true differences between controls and patients even if the mean thickness of the clusters is not statistically different? Or should i just consider the clusters that have statistically different mean thickness between groups?
Many thanks
2010/8/3 Douglas N Greve greve@nmr.mgh.harvard.edu
It might be, it depends on how you defined the sign of the voxel-wise threshold. If you used "abs", then it is certainly possible.
doyug
Joana Braga Pereira wrote:
Dear all,
I would like to ask you whether there should be an exact correspondence between vertex-wise thickness and mean thickness of clusters.
I'll explain you my problem. I did an analysis with 2 groups and 2 covariates. Then i run the mri_glmfit script and found clusters of greater thickness in controls compared to patients. When i extracted the mean thickness of these clusters and did a two sample t test i observed that the controls had greater thickness than patients, however not statistically significant, even after including in this analysis the 2 covariates.
Is this normal? How does freesurfer uses the covariates when performing an analysis?
Many thanks
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: www.nmr.mgh.harvard.edu/facility/filedrop/index.html
It depends on the test that you want to make. By using abs, your hypothesis is that there is a difference between the groups but you don't care about the direction of the difference. The mean thickness test is a different hypothesis, so it's not too surprising that you are getting different results. Both can be accurate.
doug
Joana Braga Pereira wrote:
Dear Doug,
Thanks for the reply.
Would you recommend using pos or neg sign, instead of abs for a group comparison?
Are my vertex-wise results showing true differences between controls and patients even if the mean thickness of the clusters is not statistically different? Or should i just consider the clusters that have statistically different mean thickness between groups?
Many thanks
2010/8/3 Douglas N Greve <greve@nmr.mgh.harvard.edu mailto:greve@nmr.mgh.harvard.edu>
It might be, it depends on how you defined the sign of the voxel-wise threshold. If you used "abs", then it is certainly possible. doyug Joana Braga Pereira wrote: Dear all, I would like to ask you whether there should be an exact correspondence between vertex-wise thickness and mean thickness of clusters. I'll explain you my problem. I did an analysis with 2 groups and 2 covariates. Then i run the mri_glmfit script and found clusters of greater thickness in controls compared to patients. When i extracted the mean thickness of these clusters and did a two sample t test i observed that the controls had greater thickness than patients, however not statistically significant, even after including in this analysis the 2 covariates. Is this normal? How does freesurfer uses the covariates when performing an analysis? Many thanks ------------------------------------------------------------------------ _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer -- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu <mailto:greve@nmr.mgh.harvard.edu> Phone Number: 617-724-2358 Fax: 617-726-7422 Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting <http://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting> FileDrop: www.nmr.mgh.harvard.edu/facility/filedrop/index.html <http://www.nmr.mgh.harvard.edu/facility/filedrop/index.html>
freesurfer@nmr.mgh.harvard.edu
-
Douglas N Greve -
Joana Braga Pereira