Hi,
Yes, significant results for contrast 1 (in tfce_corrp stat1) show voxels where FA in controls is greater than FA in patients.
All the best,
Mark
On 26 Sep 2013, at 10:27, Hamzah Hamid Baagil <[log in to unmask]> wrote:
> Dear FSL experts,
>
>
> I am confused about the interpretation of tfce_corrp_tstat1 and tstat 2.
> My design contrast looks like this:
>
>
> Contrast1: Controls > Patients
> Contrast2: Patients > Controls
>
>
> So if I understand correctly the corrected tfce_corrp stat1 shows cluster where controls display higher value; meaning Patients displaying lower FA value than controls and vice versa.
>
>
> I would appreciate any help.
>
>
> Thank you,
> Hamzah
>
> ----- Ursprüngliche Nachricht -----
> Von: "Wang, Yingying" <[log in to unmask]>
> Datum: Dienstag, 10. September 2013, 20:03
> Betreff: Re: [FSL] TBSS covariate
> An: [log in to unmask]
>
> > Hi Hamzah,
> >
> > Covariates are added to taken out the variance caused by these
> > covariates. So you can conclude difference in FA between
> > your patient group and control group is not due to age or performance.
> >
> > Example:
> > Your design.mat file:
> > /NumWaves is your column number.
> > /NumPoints is your row number.
> > Below it is a simple example with 3 patients and 3 controls,
> > column 3 is your age (demeaned) column 4 is your response
> > time (demeaned), about if you need to demean, please see
> > the link: http://mumford.fmripower.org/mean_centering/
> >
> > design.mat
> > /NumWaves 4
> > /NumPoints 6
> > /Matrix
> > 1 0 -9.904762 -1.0199608534
> > 1 0 -3.904762 0.4074364068
> > 1 0 1.095238 -1.4254403055
> > 0 1 -1.904762 -1.9898238671
> > 0 1 -6.904762 1.6485322973
> > 0 1 0.095238 0.8786692836
> >
> >
> > Then, design contrast file:
> > /ContrastName1 "name your contrast"
> >
> > Example:
> > design.con
> > /ContrastName1 "patient_vs_control"
> > /ContrastName2 "control_vs_patient"
> > /NumWaves 4
> > /NumContrasts 2
> > /Matrix
> > 1 -1 0 0
> > -1 1 0 0
> >
> >
> > Command:
> > randomise -i all_FA_skeletonised -o tbss -m
> > mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2
> >
> > see: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/TBSS/UserGuide
> >
> >
> > Hope this is helpful to you.
> >
> > -----Original Message-----
> > From: FSL - FMRIB's Software Library [mailto:[log in to unmask]]
> > On Behalf Of Hamzah Baagil
> > Sent: Tuesday, September 10, 2013 8:45 AM
> > To: [log in to unmask]
> > Subject: [FSL] TBSS covariate
> >
> > Dear FSL experts,
> >
> > I am trying to add a covariate (Age or Reaction time) for my
> > TBSS analysis. In my analysis, I am using a group of patients
> > compare to controls. I have tried using the GLM for the design
> > matrix, but I have difficulties adding covariate to my analysis
> > and interpreting the results.
> >
> > I would be very thankful for any help.
> >
> > Kind Regards,
> > Hamzah Baagil
|