The DTI data I have when displayed using SPM2 have '-' x voxel sign. After
calculating the FA map (using FSL3.2) and normalising to the standard space
the normalised images have - x voxel size. However, during the various steps
involved in the processing the sign of the x voxel size is not always '-' It
keeps changing. This worries me.
This is what I have done:
The B0 image when skull stripped using fsl3.2, the B0_brain image has
positive x voxel size when displayed using SPM2. (skull stripping has
changed the x voxel size sign from - to +).
When the FA map, which was generated using fsl3.2 with B0_brain as one of
the inputs, was displayed using SPM2, the x voxel size is positive as in
B0_brain image.
Next I normalised the FA maps to the standard space(ss) by first
normalising the B0_brain image to the group B0 template (which is in ss and
has '-' x voxel size) and then using write normalise to transform FA map to
the ss (normalisation is done with SPM2). The transformed FA has '-' x voxel
size when displayed using spm2. (normalisation procedure has changed the
sign of x voxel size back to '-')
My questions are :
1. What does the 'sign' of the x voxel size means ? and how is it determined ?
2. Because the final normalised image has '-' x voxel size, and the input
data to the normalisation process have '+' x voxel sign, during the
normalisation process, does B0_brain get flipped to match the ss template
which has '-' x voxel size ?
3. if there is a flipping happening at which stage does the flipping happens
? is it done before determining the parameters for normalisation, so that
the 'real left/right' of B0_brain is matched to the left/right of the template ?
I thank in advance to who ever respond to my queries.
Anusha
|