Hello
This question is about using 3Dclustsim for analysis of DTI measures.
Just to give you some background, the raw diffusion data for this project was acquired at 2mm and were processed using TORTOISE. As you can imagine, from the raw data to the final FA, MD, AD, RD maps that are being fed to 3dLME there are several points along the processing pipeline when some data interpolation is performed on the raw diffusion data from which the various DTI measures are extracted. For example, during diff prep the data are upsampled to 1mm and when we do the non-linear tensor fitting some interpolation is again introduced in the data. Furthermore, we use an in-house tensor registration method, which uses smoothing functions from ANTS in the process of creating subject-specific templates (because this is a within subject design) and study templates and bring the native DT volume to the study template space by combining the warps. The final DTI maps (i.e FA, MD etc) that are fed into 3dLME are thus in a template space.
In light of all these steps that the data has been put through, I was wondering if there are any recommendations as to what parameters I should use for 3dClustSim
Just for starters I tried the following command

3dClustSim -nxyz 192 236 171 -dxyz 1 1 1 -mask /misc/data57/thomascp/TDASP_DWI/Analysis/Statistics/SuperTempmask_TRMask_Erode1.nii
Here SuperTempmask_TRMask_Erode1.nii was created from thresholding the template TRACE (i.e. sum of the 3 diffusivities) volume at < 3000 (to remove CSF)

My question is: what smoothing level should I use? In brain parenchyma TRACE is mostly flat, while FA is not so are there any special consideration that one should bear in mind for these different types of DTI maps?

Looking forward to hearing your thoughts on this

Best regards

Cibu Thomas