AFNI Message Board

History of AFNI updates  

August 23, 2016 10:04AM
Hi Emily,

The 3dSkullStrip result is usually just used for registration,
and there are 2 main registration questions that should
always be answered (whether or not there is a concern
from 3dSkullStrip):

1. Are the EPI and anat well registered?

2. Are the anat and template well registered?

If there are extra or missing bits in the stripped dataset,
but the registration (notably in those regions) is good,
then there is no problem.

If the EPI/anat registration is affected by imperfections
from 3dSkullStrip, then it is likely there is a bigger
problem of the EPI not having sufficient contrast. In
such a scenario, consider using a pre-magnetization steady
state volume for anat/EPI registration.

If the anat/template registration is adversely affected
by the skull stripping, that is harder to solve. Peter's
suggestions of running 3dUnfize or even falling back to
Freesurfer are good.

The script to use NL registration to help skull-stripping
might indeed help the stripping, but it seems to me that
it would only really help if registration already worked
well. And in that case, it may not actually be necessary
(because it relies on a successful registration, which is
the assumed goal).

So anyway, focus first on whether the alignment looks good.
If it does, then it is likely that better skull stripping
is a useless concern.

- rick
Subject Author Posted

Skull Stripping Attachments

Emily August 22, 2016 01:06PM

Re: Skull Stripping

Peter Molfese August 22, 2016 02:44PM

Re: Skull Stripping

rick reynolds August 23, 2016 10:04AM

Sorry, only registered users may post in this forum.

Click here to login