Hi Derin,
this is how I see it:
The base might need edits and if it does, they should be made. As you say, the surfaces of the longitudinal runs are initialized with the surface of the base, so, if it is not correct, this will pollute all time points.
Another reason to edit the base is that wm edits (255) and deletes (1) are copied from the base to all time points, as you mentioned.
Usually this should be sufficient, but maybe you need further edits in a specific longitudinal run, never happened to me.
The old longitudinal processing stream really did not have any cross sectional runs. Instead of the base they processed TP1 (introducing a bias) and copied the edits from there to the longitudinals. We now copy it from the base.
Hope that helps.
Nick mentioned that edits are copied from the cross sectionals, but I think they only come from the base. At least that is what I think was done and therefore put it on the wiki. Maybe Nick can verify what really goes on in recon-all as he programmed that part?
It might make sense to edit the cross sectionals anyway because: - you then get more reliable cross sectional results (to compare to etc). - the cross sectional asegs are 'fused' (probabilistic voting), to initialize the labeling in the longitudinal runs, so more accuracy in the cross sectionals cannot hurt. However I never really found this to be necessary, but my data was usually high quality.
Best, Martin
On Fri, 2010-02-12 at 13:51 -0600, Derin Cobia wrote:
Bumping this back up to the list. Could someone please address?
To be more concise, during longitudinal processing, where are wm/gm edits made? On the cross-sectional runs, to the base, or on the .long. runs themselves? Thanks.
-Derin
Begin forwarded message:
From: Derin Cobia d-cobia@northwestern.edu
Date: February 1, 2010 3:54:46 PM CST
To: "freesurfer@nmr.mgh.harvard.edu Help" freesurfer@nmr.mgh.harvard.edu
Subject: Longitudinal processing clarification
(Questions for all, but maybe more directed towards Martin)
I have some questions about manual edits in regard to the longitudinal stream. Based on the current info on the wiki page, it appears that no edits are pushed forward from the cross-sectional (initial) runs of the subject data, as any information on the *.long.* runs come from the template. However, I've noticed that WM edits and surfs from the template are used for the *.long.* processing. In essence, my question is when/where should manual edits (i.e., control points, WM, pial edits) occur when conducting longitudinal processing in 4.5? It seems I remember that in 4.3 edits in the cross-sectional work would be pushed forward to the *.long.* images, but such is not the case anymore (?). Are edits to occur in the template (i.e., 'base') now, or in the *.long.* runs? Thanks.
-Derin
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer