Hi,
I was thinking about SIENA, how it deals with leftovers by BET, say if a
part of medulla was not completely removed by in the both volumes. If two
time-points of the same subject are compared, this may be handled by
logical OR of the coregistered BET masks. However, if these parts also
pass the OR, then they may bias percentage-brain-volume-change, by making
it relatively smaller than it actually is. If the BET leftovers are
different from subject-to-subject, the PBVCs would have different bias.
I was thinking whether intermediate registration to a template brain could
help, if the template has precisely defined brain (and its mask), i.e. no
eyes, no brainstem, ...
Thanks for explanation.
Martin
|