Thanks Bruce. I have looked at that part of the wiki but I was wondering whether the screenshots at all suggest that all artefacts are likely to be the result of the same problem, and that e.g. re-running recon for all subjects with a watershed>25 and with the -no-wsgcaatlas flag might be the thing to do in this case, rather than manually editing slices in tkmedit.
So after making these corrections to all affected subjects, before proceeding to analyses, is a visual inspection in freeview of the brain.mgz of all subjects enough? It seems like some of these skull-stripping errors are quite subtle and can be easily missed (as I have, in fact, when I checked the output after my first recon-all).
the T1.mgz is intensity corrected and conformed (the orig.mgz is just conformed).
So are both derived from 001.mgz? Otherwise, what is the difference between 001.mgz and orig.mgz?