You are fast, Gwenaëlle. Thank you!
Yes I have played with the thresholds...and they don't seem to change things very much.
Unless I am not setting them right. When I set the minimum at O and the max at .95 things look pretty much the same...
May be I don't know where to set the thresholds ? :-)
Uta
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Gwenaëlle DOUAUD
Sent: Friday, August 20, 2010 14:31
To: [log in to unmask]
Subject: [FSL] Re : [FSL] TBSS contrast
Hi Uta,
> I have conducted TBSS on 9 patients and their age and
> gender matched controls. I have then used the
> randomise tool to obtain voxelwise statistics on the
> skeletonised FA data.
>
> Contrast 1 gives me controls>patients, see yellow/red.
> Contrast 2 gives me controls<patients, see blue/light
> blue.
>
> The image is at the address below as I could not get it
> small enough to attach it to this email.
>
> http://www.admanltd.com/facomparison.jpg
>
> All are overlaid on the FA skeleton (see green)
>
> Although my patients suffer from a chronic condition, it is
> NOT a degenerative disease, so I am interested in both
> contrasts.
>
> I was very surprised (disappointed) when I looked at
> Contrast 2. What could have happened that there are
> sooo many regions in the patients (look at blue/light blue)
> that are "increased" relative to the controls?
>
> Any suggestions are very welcome,
>
> I could have made a mistake, I am an FSL novice, subject
> numbers? I am fishing.....
From your snapshot, it would seem that you have *not* thresholded your tfce_corrp images at 0.95, so you are *not* looking at significant results. Do you still have patients>controls results after looking at corrected p values<0.05?
Cheers,
Gwenaëlle
--------------------------------------------------------------------
Gwenaëlle Douaud, PhD
FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington OX3 9DU Oxford UK
Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717
www.fmrib.ox.ac.uk/~douaud
--------------------------------------------------------------------
|