External Email - Use Caution        

Hi Bram,

Thanks very much for the detailed explanation. I understand that I have to edit the wm first, and only work on the brain.finalsurfs if wm edit does not help. 

I tried the command you suggested (recon-all -s <subject_id> -white -smooth2 -inflate2 -curvHK -curvstats -autorecon3), with the new brain.finalsurfs and wm files, but the results were not ok (same as I got before). Instead I had to run (recon-all -autorecon2-wm -autorecon3 -subjid <subject>) to get the similar results as you showed in the last email. I don't understand why we get different results. 

Best wishes,

Xiaoqian









On Mon, 21 Oct 2019 at 16:33, Diamond, Bram Ryder <BRDIAMOND@mgh.harvard.edu> wrote:
Hi Xiaoquian,

That's correct -- although I did it in a slightly different order because I wasn't sure what was causing your issue. I first edited the wm.mgz using the 'recon edit' option (changing wm lesion voxels to 255) in FreeView, because I was hoping that would fix the surfaces. When that didn't work, I instead edited the brain.finalsurfs.mgz with the 'voxel edit' tool and changed the voxel values within the wm lesion to 110. As you can see in the attached images, that seemed to do the trick. The latest run-through applied both the wm and brain.finalsurfs edits -- but it's likely that the surfaces would have improve with only the brain.finalsurfs edits. You can test that out for yourself.

Instead of running recon-all between each test, I used the 'mris_make_surfaces' command exactly how recon-all would call it, but only for the rh (the hemisphere with the lesion and manual edits). The command can be found in the ReconAllTableStableV6.0 table: mris_make_surfaces -aseg ../mri/aseg.presurf -whiteonly -noaparc -mgz -T1 brain.finalsurfs <subjid> ?h . Once I saw that the new edits worked, I ran recon-all with the remainder of autorecon2 (-white -smooth2 -inflate2 -curvHK -curvstats) and all of autorecon3 (-autorecon3).

If you are not seeing the same surface improvements, make sure you've replaced your wm.mgz and brain.finalsurfs.mgz with the edited version I sent you last week. You can save copies of your original files, but rename them wm_old.mgz and brain.finalsurfs_old.mgz so that FreeSurfer doesn't accidentally apply them to your new recon-all.

Going forward, if you have a similar issue, I would first suggest editing the wm.mgz. If that doesn't work, move to the brain.finalsurfs.mgz.

Let me know if you're still having trouble or would like more detail.

Best,
Bram

___________________
Bram R. Diamond, BSc
Sr. Clinical Research Coordinator
Laboratory for Computational Neuroimaging (LCN)
Laboratory for NeuroImaging of Coma and Consciousness (NICC)
Massachusetts General Hospital
(p): 617-726-6598

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Xiaoqian Yan <yanxqpsy@gmail.com>
Sent: Friday, October 18, 2019 4:17 PM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] imperfect segmentation in the occipital lobe in epilepsy patient
 

        External Email - Use Caution        

Dear Bram, 

Thanks very much for your help. If I am understanding correctly, you first filled the lesions in the wm (brain.finalsurfs.mgz) in freeview with 'voxel edit', brush value = 110. Then you recreated a new wm.mgz, with which commands? I might run into similar situations in the future, as I am working on a project with epilepsy patients that will have depth electrodes implanted in their brain (causing thermal lesions). When that happens again, I want to be able to fix that by myself. 

Thanks again, best wishes

Xiaoqian

On Fri, 18 Oct 2019 at 20:48, Diamond, Bram Ryder <BRDIAMOND@mgh.harvard.edu> wrote:
Hi Xiaoqian,

We don't typically recommend this, but it looks like you will need to edit your brain.finalsurfs.mgz to account for the wm lesion. I tested this out by filling in the lesion with 110 and recreating the wm surface -- it seems to work quite well.

Once that is done, you can run the following command to finish your recon-all. I'm running it now to make sure the output is okay:
recon-all -s <subject_id> -white -smooth2 -inflate2 -curvHK -curvstats -autorecon3

I also made some edits to the wm.mgz, but I'm not sure they were necessary. I'll send you copies of the edited files in a private email.

Best,
Bram

___________________
Bram R. Diamond, BSc
Sr. Clinical Research Coordinator
Laboratory for Computational Neuroimaging (LCN)
Laboratory for NeuroImaging of Coma and Consciousness (NICC)
Massachusetts General Hospital
(p): 617-726-6598

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Xiaoqian Yan <yanxqpsy@gmail.com>
Sent: Wednesday, October 16, 2019 12:11 PM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] imperfect segmentation in the occipital lobe in epilepsy patient
 

        External Email - Use Caution        

Dear Bruce,

Thanks for the reply. I have uploaded my data through the Martinos Center FileDrop, here is the link:
+ Data_Xiaoqian.zip (169.46 MiB) <http://gate.nmr.mgh.harvard.edu/filedrop2/?p=6wbctqzrzh1>

I explained my problems in that link, but still copy and past here:

'' I am having troubles doing a good segmentation in the right occipital lobe in a patient's brain. There is a thermal lesion in that region, a black whole, that could make the segmentations go wrong. I tried to edit the wm, but the results were not ideal. 

The thermal lesion can be clearly seen from the axial slices 108 - 115, e.g. in RAS coordinate [40,-50,-4] on the axial slice 112. The bad segmentation (wm and pial surface) can be seen from the coronal slices 54 - 62, e.g. in coordinate [42, -52, -10] on the coronal slice 57. The wrong segmentation made it impossible to overlap the functional activations in the OFA (occipital face area) region on the surface.

In my uploaded data directory, I already edited the wm and brainmask, and I only reran 'recon-all -autorecon2-wm -subjid name' to save time. ''

Thanks in advance, and best wishes,

Xiaoqian





On Wed, 16 Oct 2019 at 15:07, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote:
Hi Xiaoqian

it's pretty hard to tell what's going on from these images. If you tar/gzip
your whole subject dir and upload it with an email telling us the specific
voxel coords that you want us to examine

cheers
Bruce


On Wed, 16 Oct 2019, Xiaoqian Yan wrote:

>
>         External Email - Use Caution        
>
> Dear Douglas,
> Thanks for your reply. I double checked my results. It was true that the
> rh.orig.nofix and rh.orig did not overlap over the region that I was
> interested in. I re-edit the wm, and only ran 'recon-all -autorecon2-wm
> -subjid patientname' to check the results, but the rh.orig.nofix and rh.orig
> were still not overlapped. Do you mind helping me check my wm edits? For me,
> the thermal lesion is a black hole that big enough to confuse the
> segmentation. Maybe there is a way to mark that region or add a label,
> before I run the recon-all?
>
> P.S. in the attached figures, pial is in green, rh.orig.nofix in red, and
> rh.orig in yellow.
>
> Thanks and have a nice day,
>
> Xiaoqian
>
>
>
> On Fri, 11 Oct 2019 at 00:47, Greve, Douglas N.,Ph.D.
> <DGREVE@mgh.harvard.edu> wrote:
>       I think that the wm edits should have worked. Can you load the
>       rh.orig.nofix and rh.orig to see if they encompass your edits?
>       If not, then maybe your did not re-run it properly.
>
>       On 10/8/2019 12:22 PM, Xiaoqian Yan wrote:
>
>               External Email - Use Caution        
>
>       Dear Freesurfer experts,
> I am trying to do the segmentation on a patient's T1 image. I
> did not get error messages running the 'recon-all' (recon-all -i
> T1.nii -subjid PatientName -all), but the segmentation over the
> right occipital lobe was imperfect, which caused failures to map
> the functional activations in this region onto the surface. 
>
> I tried to manually edit the wm.mgz several times, but the
> output did not change a lot (please see the attached image). The
> patient had depth EEG electrodes implanted in this region
> before, so there is a thermal lesion which I thought could have
> caused the bad segmentation.  Do you have any ideas about what I
> can do next, or it is possible for you to help me editing the
> data?
>
> Thanks in advance, best wishes
>
> Xiaoqian
>
> _______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>
>
> _______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>
>
>_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer