Hi Bruce,
I have never used expert options in Freesurfer. I googled "freesurfer expert options" and this link came up ( http://ftp.nmr.mgh.harvard.edu/fswiki/recon-all#Arguments), but I do not see how to specify ranges.
Is it a flag after recon-all? Could you show me an example of how to set gm T1.mgz values?
Thank you very much!
Victor
On Tue, Aug 13, 2013 at 3:55 PM, Bruce Fischl fischl@nmr.mgh.harvard.eduwrote:
Hi Victor
yes, look at the expert options. You can set the range of allowable wm and gm values (and csf for that matter).
cheers Bruce
On Tue, 13 Aug 2013, Victor Kovac wrote:
Hi Bruce,
I was using v5.1 but switched to v5.3 for these troublesome cases. The biggest issue is that this is a longitudinal study and these scans are from patients' 4th visits. Looking in the asegstats tables from all 4 visits, it would appear that the patients having this issue have had a ~20% increase in cortical WM since their 3rd visit last year, after showing a steady decline previously.
There isn't great GM/WM contrast, so I added control points to this subject. That reset the T1.mgz value of voxels that I would consider to be WM to 109 or higher, yet I still see nearly all GM with T1.mgz values in the 80's and 90's segmented as WM, which is why I am confused.
From what I can see, for this brain (MPS3008-4) only voxels with a T1.mgz of less than 80 are being segmented as GM. Is there a way for me to manually plug in the voxel intensity ranges for GM and WM and re-create the surfaces?
Thank you very much for your help with this!
Victor
On Tue, Aug 13, 2013 at 3:29 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu
wrote: never mind, I see the problem. I think this is due to the lack of gray/white contrast in this subject. I'll try some control points, but it may be that it will need some expert opts to specify the white and pial intensity ranges On Mon, 12 Aug 2013, Victor Kovac wrote:
Dear Bruce, I have uploaded the subject with unsegmented ventricles (MPS3025-4) via File Drop. This subject has an artifact from a ventricular shunt in his left hemisphere, and we are only using data from his right hemisphere. I have also uploaded a subject (image attached) whose cortical GM is being segmented as WM, and whose pial boundary is extended past the cortex. This is a recurrent issue with the atrophied brains I am working with. I added control points to bright white matter voxels in order to re-threshold WM voxel intensity and hoped that re-running autorecon-cp -autorecon3 would fix the issue, but it seems to have done nothing. Could you take a look at this subject and tell me if you know there is a way to get FreeSurfer to take my re-thresholding attempts and recreate the pial and white matter boundaries? Please let me know if there was an issue with my upload, and, again, thank you for your help! Victor On Fri, Aug 9, 2013 at 2:28 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote: hmmm, you could manually correct it but if you upload that dataset I'll take a look as well On Fri, 9 Aug 2013, Victor Kovac wrote: Dear Bruce, As I have been looking through brains I have run recon-all on with v5.3, I'm noticing that there are many cases of lateral ventricles being left unsegmented in subjects with dilated ventricles. An image of one case of unsegmented ventricles is attached, where a subject has an artifact from his shunt which may be causing the error. I have manually adjusted this subject's talairach to fix its talairach failure. Should I be using the attached script "tkmedit-lv.sh" to fill the ventricles, or is there a better method? Thank you very much for your help. Victor On Fri, Aug 2, 2013 at 1:13 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote: Hi Victor, yes, 5.3 is most of what you need. I think I added a couple of control points as well, but they should be in the subject dir i sent you. cheers Bruce On Fri, 2 Aug 2013, Victor Kovac wrote: Bruce, Thank you so much for analyzing this subject -- the segmentation looks much better! Was it simply running the subject with v5.3, or did you add any flags to the recon? As you mentioned, the pial and white matter surfaces extend further than they should, but I have found adding control points to white matter near the cortical surface has helped a lot in similar cases. I have two other subjects whose ventricles were left unsegmented, but have less atrophy than the subject I sent you. I tried filling the ventricles using the attached script, then ran recon with -autorecon2-wm -autorecon3 flags, but it did not have any effect. Would you recommend just running the subject with v5.3? Thank you! Victor On Thu, Aug 1, 2013 at 10:32 AM, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote: Hi Victor the 5.3 recon finished and you can grab the subject from: http://gate.nmr.mgh.harvard.**edu/filedrop2/?p=if8d3r4<http://gate.nmr.mgh.harvard.edu/filedrop2/?p=if8d3r4> I think the ventricles are *much* better. The white surface was incorrect in anterior temporal regions due to lack of myelination, but that is expected. You might be able to improve them, but there is little enough contrast that I'm not sure how good they will be cheers Bruce On Tue, 30 Jul 2013, Victor Kovac wrote: Dear Freesurfer experts, I am running recon-all on very atrophied brains with mixed results. Attached is an image of a brain where segmentation has gone awry. There is no message indicating talairach failure, and the segmentation appears to need WM and pial edits as well as the ventricles to be filled, so I am not sure how to proceed. If you have any recommendations as to how to fix these errors, I would very much appreciate it -- these data are past their deadline, and I am at a loss. Thank you! Sincerely, Victor The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/**complianceline<http://www.partners.org/complianceline>. If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.