Hi,
The recommendations in TBSS are definitely aimed at typical human brain data, so it is quite reasonable for you to adapt the method to different data. As long as you continue to use the FA to define the skeleton and do the projection then using the b0 for registration is totally fine.
All the best,
Mark
> On 12 Nov 2017, at 07:02, Patrick <[log in to unmask]> wrote:
>
> Hi Experts
>
> I have a question regarding TBSS on rodent data. I have rewritten the TBSS scripts to work with rodent data but I had a question specifically regarding the 2nd step (tbss_2_reg).
>
> In this step I have used my b0 volumes to estimate the registration rather than the FA maps. This gives me a much stronger registration than the FA maps specifically. Then I simply apply those transforms/warps to my FA/MD maps (which are in the same space as my b0's) and continue on with TBSS. I have rewritten the scripts (tbss_2_reg removed the -FA flag, and tbss_3_postreg used my own template) to eliminate the need to warp into standard space etc. as they previously did not work well with rodent data.
>
> Do you see any issues with the statistical inferences from this approach?
>
> Thanks
> Patrick
|