1. This is based on Randy Buckner's paper showing that the tal xform gives you a pretty accurate estimate of the TIV (which is hard to compute directly from a T1 image since you can't distinguish bone from CSF).

2. I think including it as a covariate is a better approach, but perhaps others can comment. The volumes require some type of normalization/regression as otherwise you are introducing uninteresting variance (since volume is correlated with head size). Thickness is pretty much uncorellated, so you don't need to control for TIV in a thicknes study.

cheers Bruce

On Wed, 8 Jun 2016, Otília wrote:

Greetings,

I have two questions regarding FS volume measurements.

I intend to use the volume information of .stats files but I got some doubts regarding the methods used.

1. I read that eTIV estimation is based on the TAL transformation, but what

about the volumes of cortical and subcortical areas? Is it a simple voxel count?

2) In some studies the authors normalize the volumes to eTIV. In others the authors don't do any normalization. Some of them use it as co-variate. What's your opinion regarding this matter?

Thank you in advance!

Best regards,