Hi Eddy,
I see now.
You can reconstruct a tensor from its decomposition using the
formula: Tensor = sum( Li * Vi * Vi')
This can be implemented easily using several calls to fslmaths (after
splitting the vectors into x,y,z components using fslsplit).
Note that you cannot use tbss_non_FA directly on vector images such as
V1 etc. You will need to edit the tbss_non_FA script, and use vecreg
instead of applywarp. Vecreg will apply a local rotation to the
vectors that is induced by the warpfield.
Cheers,
Saad.
On 10 Mar 2009, at 10:59, Eddy D.L. wrote:
> Hi Saad.
> sorry, I try to explain better.
> I start from a number of patients DTI study (file .nii.gz), use the
> technique TBSS then I use the tbss_non_FA to retrieve the
> eigenvalues and
> working separately with the eigenvectors. Now I have L1, L2, L3, V1,
> V2, V3
> and I would like rebuild final DTI.
> I want a single image ( for example, template.nii.gz ), where, each
> voxel
> contains a matrix 3x3 (tensor).
>
> Thanks
>
> Eddy
>
>
>
>> Hi Eddy,
>>
>> I am not sure I fully understand what you are saying here. What do
>> you
>> mean by DTI "complete"?
>> In FSL, the program that generates L1,L2,L3,V1,V2,V3 is DTIFIT, not
>> TBSS.
>> TBSS uses the output of DTIFIT in order to do multi-subject voxel-
>> wise
>> analyses.
>
>> In DTIFIT, if you want to output the tensor, you need a call like
>> this:
>>
>>> dtifit -k data -o dti -r bvecs -b bvals -m nodif_brain_mask --
>> save_tensor
>>
>>
>> Cheers,
>> Saad.
>
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|