Dear all,
I am trying to use tbss v1.2 on FA volumes, with 2 groups: patients (n=3)
vs controls (n=43). The patients have no or little tissue both medially
and laterally in the anterior parts of the left temporal lobe, and no or
little tissue medially in posterior parts of the same lobe.
I have completed the tbss stream without obvious problems or
misalignments, but I find it somewhat difficult to judge if the nonlinear
registration has worked properly in the patients, especially in the areas
surrounding the parts with no or little tissue.
In trying to get around the problem of judging the result of the automatic
process I have also tried to do the 'tbss_2_reg' with individual commands
and not the tbss_2 script, with the --inmask options in fnirt masking the
area with no or little tissue in the patients. My assumption has been that
'tbss_2_reg -T' equals the following three processing steps (of course
with the exception of the '--inmask=binary_templobemask.nii.gz'):
1. flirt -ref ${FSLDIR}/data/standard/FMRIB58_FA_1mm -in subject_FA.nii.gz
-omat subject_FA_to_target.mat
2. fnirt --in=subject_FA.nii.gz --inmask=binary_templobemask.nii.gz
--subsamp=8,4,2,2 --infwhm=12,6,2,2 --reffwhm=12,6,2,2
--lambda=300,75,30,30 --estint=1,1,1,0 --aff=subject_FA_to_target.mat
--cout=subject_FA_to_target_warp --config=FA_2_FMRIB58_1mm
3. applywarp --in=subject_FA.nii.gz --out=subject_FA_to_target
--ref=${FSLDIR}/data/standard/FMRIB58_FA_1mm
--warp=subject_FA_to_target_warp
In addition to the files from tbss_1 this seems to create all the files I
need in order to proceed to tbss_3, but still:
Q1: Is this at all feasible/adviceable?
If so:
Q2: Is my initial assumption of equality valid?
Q3: If not, what do I need to add/subtract in order to do the tbss_2_reg
with a mask?
Comparing the subject_FA_to_target.nii.gz output from the tbss_2_reg and
the same file from the command lines above there is no difference,
therefore:
Q3: Am I using the --inmask option wrong or what else am I missing? The
'binary_templobemask.nii.gz' is "a "hole of zeros" in a binary image with
ones" as it says on the fnirt web page, in the same space as the --in
image.
Thank you in advance!
Best,
Håkon Grydeland
|