Hi,
The analysis you are proposing here (looking at the width of the tracts) is not something that TBSS does, and so you are right that this information is not used (or lost). However, I doubt that the registrations are sufficient robust and accurate enough to handle subtle geometric changes like this without being influenced by artefacts, partial volume effects and other sources of variability in the diffusion (possibly biological and possibly not). We are working on a more diffusion-specific tool for registration, but even if we had that it would be difficult to interpret the changes from such a method without some fairly careful validation work. Therefore, I would not recommend doing anything like this with the current tools, but would instead stick to TBSS analysis of FA (or other diffusion parameters) or VBM-style analysis of GM density.
All the best,
Mark
On 29 Aug 2012, at 12:47, SUBSCRIBE FSL Jay <[log in to unmask]> wrote:
> Hello Jesper, Mark,
>
> Yes, I understand that, when we are looking at the degree of anisotropy, especially on the skeleton, the FA value does not change in the warped image.
>
> But, what would it be in case of subject_1(constant white matter FA value = 0.7) who has twice as thick white matter bundle as subject_2(constant white matter FA value = 0.7) all over the brain. I assume that, when these two subjects FA data are projected on the mean_skeleton, the degree of anisotropy would not show any difference between the two subjects(since both subjects warped white matter FA value on the skeleton should be 0.7).
>
> If that is the case, the information about subject_1 having thicker white matter bundle would be lost. If i wanted the degree of anisotropy to show me not only the difference in micro structural anisotropy but also the volumetric difference in the thickness of the white matter bundle, then would it make sense to perform volume correction also?
>
> BW
> Jay
>
|