On 11 Mar 2009, at 10:45, Eddy D.L. wrote:
> Hi Saad,
>
>> If you want the average of L1 on the skeleton, you need to use
>> all_L1_skeletonised, NOT all_L1
>> This is because all_L1_skeletonised is the value of L1 projected onto
>> the skeleton.
>> In order to calculate the mean of all_L1_skeletonised, you only
>> need a
>> single call to fslmaths:
>>
>> fslmaths all_L1_skeletonised -Tmean mean_L1_skeleton
>>
>> I guess your three lines above were used to exclude negative values
>> from L1, that fine!
>
> I saw that the process tbss_non_Fa give me out all_L1.nii.gz, this
> image has
> three volumes, but image of eigenvalue L1 should have a single
> volume, or
> wrong? So I copied those lines from the file bach TBSS. I do not
> know if I
> have explained well?
If all_L1 has three volumes, that means you have three subjects!
How many volumes does all_FA.nii.gz have? This should be the same as
all_L1, otherwise something weird happened.
> Then I use L1 in subsequent steps, to reconstruct the tensor.
> So is correct use: fslmaths all_L1_skeletonised -Tmean
> mean_L1_skeleton?
Yes.
>
>
>
>> fslmaths V1_subject2 -mul V1_subject1 -add V1_dot
>> fslmaths V1_dot -abs -div V1_dot V1_sign
>> fslmaths V1_subject2 -mul V1_sign V1_subject2_reoriented
>> And then add all the V1_subjectn_reoriented to get the mean
>
> It is now clearer thanks
> For the last step is correct, with n subjects:
> fslmaths V1_subject1 -add V1_subject2_reoriented (..) -add
> V1_subjectn_reoriented V1
> where V1 is the final result! is right?
that's right, except that you need to add an extra option to divide by
the number of subjects:
fslmaths V1_sub1 - add V1_sub2 (...) -add V1_subn -div n V1_average
>> fslsplit V1 V1
>> fslsplit V2 V2
>> fslsplit V3 V3
>> fslmaths L1 -mul V10000 -mul V10000 tmp1
>> fslmaths L2 -mul V20000 -mul V20000 tmp2
>> fslmaths L3 -mul V30000 -mul V30000 tmp3
>> fslmaths tmp1 -add tmp2 -add tmp3 tensor11
>
> When I have all the components T11, T12 .. Can I use a simple "-add"
> to
> obtain a single image? that is an image with nine volumes, as in the
> last
> passage of eigenvectors
No, you need to use fslmerge in order to concatenate the images onto a
single 4D image:
fslmerge -t avg_tensor T11 T12 T13 T22 T23 T33
>
>
> thank you very much
>
> Eddy
>
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|