Hi Donatella,
> Date: Mardi 8 Juillet 2008, 9h57
> Hi,
> I'd like to ask you a question about TBSS outputs.
> The randomise tool generate the files:
> 1)tbss_tstat1 with data that give control>patient (i.e.
> voxels with FA values for control group > FA values for
> patient group, isn't it?)
Yes, if you have entered your controls first in the design matrix and went for a 1 -1 contrast.
> 2)tbss_tstat2, the contrary
Yes.
> 3)tbss_maxc_tstat1 with data that give control>patient
> with the
> correxted p-value (i.e voxels with FA values for control
> group > FA values for patient group
> with the corrected p-value using the Random Field Theory
> for the evaluation
> of the Family-wise errors?)
No, the FWE 1-(p-values) are given in the tbss_max_tstat1, not maxc. Maxc is giving you the 1-(p-values) corrected using a cluster-based approach, thresholded according to the value you have entered after the -c option (please look at the manual, http://www.fmrib.ox.ac.uk/fsl/randomise/index.html).
Hope this helps,
Gwenaelle
_____________________________________________________________________________
Envoyez avec Yahoo! Mail. Une boite mail plus intelligente http://mail.yahoo.fr
