Dear FSL experts,
I have run the tbss_sym script on my DTI data as suggested in the TBSS guide. After testing for significant asymmetries in diffusion characterisics, I was wondering whether I could investigate how this asymmetries correlate with some cognitive score I have for my subjects. However, if I set up the simple GLM to test for correlation with the following design matrix and contrast (note that I have already demeaned my EV):
/NumWaves 1
/NumPoints 35
/PPheights 2.671244e+00
/Matrix
-6.154478e-01
-2.434025e-01
[...]
2.056410e-01
/ContrastName1 posMix
/ContrastName2 negMix
/NumWaves 1
/NumContrasts 2
/PPheights 2.671244e+00 2.671244e+00
/RequiredEffect 2.787 2.787
/Matrix
1.000000e+00
-1.000000e+00
and call randomise with the command
randomise -i all_FA_skeletonised_left_minus_right.nii.gz -o result -m mean_FA_symmetrised_skeleton_mask_dst.nii.gz -d Mixing_contrast_design.mat -t Mixing_contrast_design.con -D --T2
i get the following warning:
"Warning: The unpermuted statistic image for the current image contains no positive values, and cannot be processed with TFCE. A blank output image will be created."
Is the idea behind this test wrong? Or does this warning come from some troubles in setting up the Glm matrix?
Alternatively I was thinking about comparing the simple correlation I find on the symmetrised skeleton directly, by looking at omologous voxels on the right/left side of the skeleton. Is this procedure right?
Thank you in advance for the support. Kindest regards,
Eleonora Mastrorilli
|