I have been working with some cortical thickness values of a control group to test some changes related to cognitive training. However, I have been hesitating about whether I have to normalize the cortical thickness values (as happens with subcortical volumes and ICV) or not. I have read a little about it and I understood that I could use the ICV or/and mean cortical thickness, but I didn't find a specific criteria to do or not this normalization. Any clue, recommendation, explanation or paper about this will be very helpful.

Thanks in advance. 

*or reposting discussion of this question.