Hello,
I am looking at a group of patients pre-post drug and placebo treatment, attempting to find regional changes in GM density (concentration) after drug treatment.
I am modelling this group using a flexible factorial design, using the un-modulated, smoothed images. I have added age as a covariate, and got some good results.
I used the IBASPM - (calculate brain volume) function to obtain GM, WM, and CSF volumes from the original T1.spgr's, and added these up to get total intracranial volume (TIV)
I then added total intracranial volume into the model as a covariate, and this significantly improved my results. I have read different things about this. Can someone please let me know if it is "statistically acceptable" to add TIV as a covariate when looking at density (i.e. un-modulated images).
best,
Anson K.
|