Hi Falk

have you tracked down the source of the defect in the wm.mgz? Sometimes very minor edits can correct even major defects. If you are having trouble with this feel free to upload the dataset and one of us will take a look Bruce On Sat, 23 Mar 2013, falk.luese@gmx.net wrote:

Hi Bruce,   well, there is an option called -correct_defect within mris_fix_topology, which is meant to correct just one defect at a time. However, after the defect is corrected one is getting a message like:   ?TERMINATING AFTER CORRECTED DEFECT   no such file or directory?   I haven?t been able to figure out how to save this or if this is even possible. Maybe I am missing the correct flag to do so. May it otherwise be possible to remove certain defects from ?h.defect_labels or something similar?   The errors of the filled.mgz result in a faulty surface reconstruction of ?h.orig.nofix and every following surface like ?h.white and ?.pial in that area. Therefore I wanted to remove these in before and, if possible, automatically.   Best, Falk   Von: Bruce Fischl Gesendet: ?22?. ?März? ?2013 ?19?:?14 An: ""Falk Lüsebrink"" falk.luese@gmx.net Cc: FreeSurfer Betreff: Re: [Freesurfer] Two questions regarding mri_fill and mris_fix_topology   Hi Falk

I don't think it's possible to fix just one defect, unless Florent included some option that I am not aware of.

I also wouldn't worry too much about the filled.mgz. Make sure that the aseg.mgz and the ?h.white and ?h.pial are accurate cheers Bruce On Fri, 22 Mar 2013, "Falk Lüsebrink" wrote:

...

Dear FreeSurfers,   I am processing some T1-weighted images of the human brain with an

isotropic resolution of .5 mm using version 5.1.

...

After mri_fill my hemispheres are separated nicely and the cerebellum is

completely removed. However, looking at the filled.mgz, posterior and

...

superior of the corpus callosum some residual of white matter can be found

in the longitudinal fissure. It does not remain if I look at the

...

conformed data, altough these leftovers are found in the wm.mgz of both

datasets. As this does not happen using conformed data I assume that the

...

cutting plane is not "broad" enough in case of higher resolution data or

something related. Is there some way to tweak this?

...

Secondly, I currently skip the -fix stage of recon-all completely while

processing this data, but I would like to correct at least some of its

...

defects. I already I have seen in the help of mris_fix_topology that it is

possible to fix just one defect and I extracted the defects and their

...

number of vertices using the defect-seg script. Is it possible to save the

resulting surface, so I am able to correct the defects I want

...

to consecutively?   Best, Falk