Doug,

I tried to replicate again v5.3 results of subject 1. I checked wm.mgz, orig.mgz, ?h.orig, ?h.pial, ?h.white and found everything identical (by using mri_diff and mris_diff) except FLAIR.mgz and ?h.pial (after FLAIRpial refinement).

The difference was small, I have traced it to the level of bbregister, namely to flirt.fsl. I cannot trace the exact version of flirt used in the original data, neither cannot check contents of xyztrans.sch, but I suspect that this is the reason of difference.

My current version of flirt reads: FLIRT version 6.0. I have uploaded the replicated subject 1 as file subj_1_v5.3_replicate_2.tar.gz to your server (fist upload attempt of file subj_1_v5.3_replicate_2.tar.gz was aborted). I have added also current xyztrans.sch.

Antonin

Doug,

did you process from scratch or did you retain edits to brainmask.mgz and wm.mgz?The -cubic should not be necessary, since recon-all (which I primarily obtained from https://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/5.3.0-patch/ has UseCubic=1 ).

I could try to replicate v5.3 again on my side.

Antonin

It is subject #1, trying to replicate in our version of 5.3 using the scripts/recon-all.local-copy ./recon-all.local-copy -all -FLAIRpial -s 1.dng.v53.local-cubic -cubicThe results were close, but they were not exact

On 5/1/17 4:13 PM, Antonin Skoch wrote: Dear Doug,

that is strange. What precisely you cannot replicate? The results with v5.3 or with v6.0? What subject from the group I uploaded you have tried? I could try to run the comparison again but I have seen the results already in many subjects and the difference between -cubic and no cubic seems profound and systematic in favor of v5.3 in the aspect of wm.mgz leak ouside brain and GM/WM contrast.

Concerning the expert options file: Despite using flag -xopts-use in recon-all in v5.3, I did not use any expert option file in the subject I have uploaded. The reason of using -xopts-use was that I was processing the subjects in batch where some of them had expert-options file with entry bbregister -init-header due to the fact that init-fsl failed in these subjects. I wanted to make my life easier by processing all of them by identical recon-all command line parameters and added -xopts-use to all subjects (even in the subjects without expert option file). I supposed that this could not do any harm.

Antonin

I can't seem to replicate your results locally, even with the recon-all you used. The one thing I'm missing is the expert options file. Can you send that to me? On 04/24/2017 12:49 PM, Antonin Skoch wrote:

Dear Doug,

the subject with leak of white surface outside brain (my first post with screenshots) is subject 1. Slice number (coronal) around 100. The subject in second post (with text below) is subject 2, slice number (coronal) also 100.

I have processed the subjects with v 6.0 (in fact dev version from feb 2017, but this is irrelevant) with -cubic -no-mprage. It looks much like v5.3, i.e. the wm.mgz and surfaces are much better, but v5.3 looks still better, at least for subject 1. I have uploaded them as file v6.0_cubic_no_mprage.tar.gz to your ftp site.

The optical difference in norm.mgz/brain.mgz between v5.3 and v6.0 with -cubic is very minor, but still there is some other thing which renders wm.mgz worse than with v5.3 for subject 1. The -cubic has profound effect, the images seem much smooth with lose of contrast without using -cubic.

Regards,

Antonin

And what slice number? On 04/24/2017 11:16 AM, Douglas N Greve wrote:

...

Anonin, of the three subjects you sent, which one is shown in these pictures?

On 04/19/2017 05:23 PM, Antonin Skoch wrote:

...

Dear experts,

I am sending just one more example to illustrate issue with white surface estimation in v6.0. See the attached screenshots: In v6.0 there seems to be insufficient contrast in brain.finalsurfs.mgz, so the white surface is leaking at three spots dramatically outwards towards pial surface. The white surface in v5.3 looks much more anatomically relevant in the same spot.

Could you please comment on how to avoid such issues in v.6.0?

Regards,

Antonin Skoch