HI Michiko,

Please see below:

On 21 August 2015 at 08:57, Michiko <[log in to unmask]> wrote:

Hi FSL experts,

Not sure if this has been discussed on another thread (tried searching but did not find any that explicitly addressed this), but is it advisable for one to always include the whole brain volume/ICV as a covariate either/both: a) in the GLM when performing randomise in TBSS analyses,

Not necessary. FA doesn't have a known, relevant relationship with brain volume or ICV.

 

b) when regressing mean FA values across the WM skeleton for individuals (using fslmeants -i all_FA_skeletonised to obtain values) against other variables (e.g. cognitive, clinical scores)?

Likewise, not needed.

 

I get that the images have been normalized to a standard space, but it does seem like common practice in a couple of papers I've read, so I'd like to understand if this practice is necessary/strongly encouraged?

There might be other reasons why one would want to include, e.g., as a potential confound for other measurements that went into the analysis. But in principle, not needed. If the sample is large, it shouldn't cause harm, though.

All the best,

Anderson

 

Thanks very much in advance.