On 12 Mar 2009, at 02:19, Angelica Hotiu wrote:

> Dear FSL experts,
> I was running TBSS analysis using DTI data acquired from  5 control  
> subjects
> and 3 subjects with mild traumatic brain injuries. All the  
> preprocessing has
> been done using FSL4.1.2 .I am  interested to identify the regions  
> where FA,
> MD and transversal diffusivity  indicate differences between two  
> groups. For
> randomisation procedure  I used first TFCE option .As a result FWE  
> corrected
> shows the results only for MD. I couldn't obtain FWE corrected for  
> multiple
> comparisson  for FA and transversal diffusivity using TFCE option.   
> Also I
> was trying randomisation using cluster-based thresholding corrected  
> for
> multiple comparisons setting different values for threshold but  I  
> couldn't
> obtain corrected values for p maps in a range [0.95,1]. Based on the
> hypothesis that  regions such as corpus callosum,  anterior  and  
> posterior
> limb of internal capsule are more vulnerable to traumatic brain  
> injuries  I
> was interested to  testing this hypothesis for spleninum of corpus  
> callosum.
> 1)I'm wondering if this is legal, to limitate TBSS analysis to some  
> specific
> region in the brain.

yes - as long as you honestly had the hypothesis before looking at the  
initial randomise results!
Just reduce the mask used in randomise to the region you care about,  
as you've done below.

> 2)If it is, could you please check if it's correct
>               draw a  mask of spleniunm of corpus callosum (mask_SCC)
>               fslmaths  mask_SCC -mas mean_FA_skeleton_mask   
> mask_SCC_new
>               randomise -i all_FA_skeletonised -o tbss  -m  
> mask_SCC_new -d
> design.mat -t design.con -n 500 - -T2 -V
> As a result of permutation, corrected p image tbss_tfce_corrp_tstat1  
> with
> values [0.95 1] indicate a reduction of FA in mild traumatic brain  
> injury
> group in spleninum of corpus callosum.

Jolly good.

> 3)I'm wondering about the validity of  the previous steps.
> If this is valid, why  this result didn't come up by using the
> mean_FA_skeleton_mask as the mask to feed into  randomise.

Because you had more voxels in the mask so the multiple comparison  
correction is more aggressive.

> 4)Another concern is related to the new version v2.1 for randomise.  
> I did
> the same analysis for the same subjects using FSL4.1.1 randomise v  
> 2.0 and I
> was able to have more results .  What is the explanation? Which one  
> is more
> recommended?

That depends on the model - but we've improved the handling of (e.g.)  
confound covariates and the interaction between EVs and contrasts wrt  
permutation - so the latest approach is more accurate, you should  
definitely use the latest version.


> I'm so sorry for the basic questions. Any suggestions will be greatly
> appreciated.
> Many thanks in advance.
> Best  Regards,
> Angelica

Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]