We are having a difficult time getting the ICV to fit correctly in our sample, but the segmentions and parcellations look accurate.  I believe that the ICV is generated from the talairach.xfm early on in the processing stream, but then further processing for segmentations/parcellations is done. 

So, does the ICV measure affect regional brain volumes in any way, or can we trust the volumetric measures to be accurate even if the ICV is not? 

Thank you,

Heather Lugar