Dear Ed,
> By means of the option -A BET produces the inner skull mask. I have
> a few questions about this mask
>
> 1.
> As it name suggests the volume of the inner skull mask is an estimate of the ICV.
> Is that correct?
If you are using both T1-weighted and T2-weighted acquisitions (with -A2) then this is possible. Otherwise I would not trust the results to be particularly good, as you cannot reliably distinguish between CSF and other structures, such as the skull and marrow, within a T1-weighted scan. BET will do the best that it can, but you really need a T2-weighted scan to reliably see the inner skull surface.
> 2.
> It frequently happens that the inner skull mask is smaller than the brain mask, i.e.,
> voxels inside the brain mask are located outside the inner skull mask. Of course,
> this is anatomically not correct and therefore I would like to ask you to modify BET
> in this respect.
> May I conclude that the algorithm to derive the brain mask is different from the
> algorithm which derives the inner skull mask?
Yes, it is a different algorithm, and it tends not to work as well unless it has both T1-weighted and T2-weighted images.
> 3.
> Related to question 2: we are frequently editing manually the brain mask. It’s a
> pitty that this modified brain mask can not be used as input to BET to improve
> the computation of the inner skull mask. Is it possible to add such an option to BET?
Not easily, and we are working on other ways of improving brain extraction.
If you are not using both T1-weighted and T2-weighted scans then try this, as that is more likely to make a difference.
All the best,
Mark
>
> Cheers,
> Ed
|