Re: [Freesurfer] longitudinal stream v 5.3 - transfer edits of the wm.mgz from cross to base?
Dear Martin,
thanks for the feedback. We had to edit cross due to the fact that
1. our dataset is involved also in cross-sectional studies 2. we are continuously acquiring timepoints but want to do some analyses with data we already have
Therefore only editing the base was not an option for us.
I was thinking (to save time spent on edits) of transferring the INTERSECTION of wm edits from all cross points to base. I think this should work, the only issue I see here is how to handle resampling from space of cross to the space of base. What do you think?
And, regarding the v6.0 is currently released, would you recommend to re-process our cross-sectional data (originally processed with v5.3) with v6.0? Would you expect any issues with cross-sectional edits made on v 5.3? Most of the edits are on the wm.mgz, some on brainmask.mgz or brain.finalsurfs.manedit.mgz.
Regards,
Antonin
Hi Antonin,
in most cases it should be sufficient to only edit wm.mgz in the base, instead of editing all time points. This is a big time saving especially if you have many time points.I don’t think merging edits from cross will work well, as there could be atrophy and lots of non-linear deformations in some of the fast progressing brains.
Best, Martin
On 23 Jan 2017, at 23:11, Antonin Skoch a...@ikem.cz wrote:
Dear experts,
I am currently starting longitudinal analysis of data I previously corrected in cross-sectional stream (FreeSurfer v5.3). The analysis comprises quite large number of subjects - approx 150 subjects, each scanned twice. All subjects are already inspected and errors corrected in cross-sectional analysis. Some of them needed quite large edits, mainly to the wm.mgz and the whole process took large time.
As I read the documentation, the wm.mgz edits in cross are not currently transferred to base. However, wm.mgz has to be correct in base in order to estimate surfaces correctly. From this it implies to me that I would need to edit all wm.mgz errors in base again. Do you think that it could be possible/reasonable to transfer some wm.mgz edits from cross to base, for example to make intersection of edits of all cross timepoints and to transfer it to the base? This could significantly reduce time spent on edits. Does it make sense?
Regards,
Antonin Skoch
Hi Antonin,
I see. The problem with the intersection would be that it could be zero in spite of several edits that move around. Still would be better than nothing. Feel free to test that (e.g. you can load that in matlab, intersect and then apply to the wm from base to see if it helps). Let me know.
About 6.0 usually edits should carry through. Test it in one or two cases to make sure it works. Also you could keep 5.3 cross and (consistently for all individuals) run 6.0 for base and long. Should also work.
Best, Martin
On 24 Jan 2017, at 23:40, Antonin Skoch ansk@ikem.cz wrote:
Dear Martin,
thanks for the feedback. We had to edit cross due to the fact that
- our dataset is involved also in cross-sectional studies
- we are continuously acquiring timepoints but want to do some analyses with data we already have
Therefore only editing the base was not an option for us.
I was thinking (to save time spent on edits) of transferring the INTERSECTION of wm edits from all cross points to base. I think this should work, the only issue I see here is how to handle resampling from space of cross to the space of base. What do you think?
And, regarding the v6.0 is currently released, would you recommend to re-process our cross-sectional data (originally processed with v5.3) with v6.0? Would you expect any issues with cross-sectional edits made on v 5.3? Most of the edits are on the wm.mgz, some on brainmask.mgz or brain.finalsurfs.manedit.mgz.
Regards,
Antonin
Hi Antonin,
in most cases it should be sufficient to only edit wm.mgz in the base, instead of editing all time points. This is a big time saving especially if you have many time points. I don’t think merging edits from cross will work well, as there could be atrophy and lots of non-linear deformations in some of the fast progressing brains.
Best, Martin
On 23 Jan 2017, at 23:11, Antonin Skoch a...@ikem.cz wrote:
Dear experts,
I am currently starting longitudinal analysis of data I previously corrected in cross-sectional stream (FreeSurfer v5.3). The analysis comprises quite large number of subjects - approx 150 subjects, each scanned twice. All subjects are already inspected and errors corrected in cross-sectional analysis. Some of them needed quite large edits, mainly to the wm.mgz and the whole process took large time.
As I read the documentation, the wm.mgz edits in cross are not currently transferred to base. However, wm.mgz has to be correct in base in order to estimate surfaces correctly. From this it implies to me that I would need to edit all wm.mgz errors in base again. Do you think that it could be possible/reasonable to transfer some wm.mgz edits from cross to base, for example to make intersection of edits of all cross timepoints and to transfer it to the base? This could significantly reduce time spent on edits. Does it make sense?
Regards,
Antonin Skoch
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Antonin Skoch -
Martin Reuter