Hi Lilla et al,
I am preparing to CVS register 30+ subjects to improve FSL probabilistic tractography alignment across subjects. First, thanks for a really cool tool! Then, I have some questions:
1. mri_cvs_register step1 is spherical registration. However, I have already run FS5.3 recon-all -all for all subjects, which includes spherical registration. Are the recon-all and mri_cvs_register spherical registrations identical? If so, will mri_cvs_register be able to use the previously computed spherical registration (with flags --step2 --step3), or will it need to compute a new spherical registration? Will mri_cvs_register overwrite any of my earlier recon-all results?
2. What are the pros and cons for using one of the following as templates, and which one would you recommend: -Randomly selected individual subject -CVS35 -CVS35_inMNI152
3. When using MNI152_1mm as template, I guess this is a 1 mm resolution image (as $FREESURFER_HOME/subjects/cvs_avg35_inMNI152/mri/T1.mgz seems to be in 1 mm resolution)? Technically, is this the cvs_avg35 image that has been CVS-registered with the MNI152_1mm image? Or is this a FLIRT-type 3D registration from cvs_avg35 to MNI152_1mm without surface information (I guess extracting the surface from MNI152 would be difficult)?
4. I run into the following error in FS stable 5.3 (source /usr/local/freesurfer/nmr-stable53-env):
mri_cvs_check --mov $SUBJECT Using the CVS template as registration target. The following files are missing: /usr/local/freesurfer/stable5_3_0/bin/mri_cvs_register.settings.txt You need to find these files or run reconall on the data in order to run mri_cvs_register.
Is this file required, or is this a bogus error?
5. Are there parallel processing versions that could be run on the cluster?
Thanks!
Tommi
--- Tommi Raij, MD, PhD TM Core Director MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging Bldg 149, 13th St Charlestown, MA 02129 U.S.A.
Hi Tommi,
I am preparing to CVS register 30+ subjects to improve FSL probabilistic tractography alignment across subjects. First, thanks for a really cool tool! Then, I have some questions:
- mri_cvs_register step1 is spherical registration. However, I have
already run FS5.3 recon-all -all for all subjects, which includes spherical registration. Are the recon-all and mri_cvs_register spherical registrations identical? If so, will mri_cvs_register be able to use the previously computed spherical registration (with flags --step2 --step3), or will it need to compute a new spherical registration? Will mri_cvs_register overwrite any of my earlier recon-all results?
The two calls are similar but not identical. In the recon call your registration target is different and the call uses a different set of parameters. So you should still call step1 in your processing and no worries, CVS does not overwrite your previous solutions.
- What are the pros and cons for using one of the following as templates,
and which one would you recommend: -Randomly selected individual subject -CVS35 -CVS35_inMNI152
That is your call and the decision will probably depend on your application or intended use of the results. Using an individual subject as a template can bias further population comparison, for example, as the subject might not be the most representative one for your population. Using the CVS35 atlas diminishes this bias and the CVS35_inMNI152 would provide you with results in the MNI152 space, so if you need to report results in that space that could be useful.
- When using MNI152_1mm as template, I guess this is a 1 mm resolution
image (as $FREESURFER_HOME/subjects/cvs_avg35_inMNI152/mri/T1.mgz seems to be in 1 mm resolution)? Technically, is this the cvs_avg35 image that has been CVS-registered with the MNI152_1mm image? Or is this a FLIRT-type 3D registration from cvs_avg35 to MNI152_1mm without surface information (I guess extracting the surface from MNI152 would be difficult)?
It is the CVS35 average transformed and reconned in the MNI152 space. It definitely has surface information, otherwise CVS would not run.
- I run into the following error in FS stable 5.3 (source
/usr/local/freesurfer/nmr-stable53-env):
mri_cvs_check --mov $SUBJECT Using the CVS template as registration target. The following files are missing: /usr/local/freesurfer/stable5_3_0/bin/mri_cvs_register.settings.txt You need to find these files or run reconall on the data in order to run mri_cvs_register.
Is this file required, or is this a bogus error?
That file is required. I will double-check why that is not in the distribution.
- Are there parallel processing versions that could be run on the cluster?
Yes. You can use --openmp to take adventage of it. The default is 1.
Let me know if you have any other questions. Lilla
Thanks Lilla!
Which MNI152 version was used as template when doing nonlinear CVS-registering from cvs_avg35 into MNI152 space? For example, was this MNI152 ICBM 2009a 1mm Nonlinear Asymmetric?
Bests, Tommi
Hi Tommi,
I am preparing to CVS register 30+ subjects to improve FSL probabilistic tractography alignment across subjects. First, thanks for a really cool tool! Then, I have some questions:
- mri_cvs_register step1 is spherical registration. However, I have
already run FS5.3 recon-all -all for all subjects, which includes spherical registration. Are the recon-all and mri_cvs_register spherical registrations identical? If so, will mri_cvs_register be able to use the previously computed spherical registration (with flags --step2 --step3), or will it need to compute a new spherical registration? Will mri_cvs_register overwrite any of my earlier recon-all results?
The two calls are similar but not identical. In the recon call your registration target is different and the call uses a different set of parameters. So you should still call step1 in your processing and no worries, CVS does not overwrite your previous solutions.
- What are the pros and cons for using one of the following as
templates, and which one would you recommend: -Randomly selected individual subject -CVS35 -CVS35_inMNI152
That is your call and the decision will probably depend on your application or intended use of the results. Using an individual subject as a template can bias further population comparison, for example, as the subject might not be the most representative one for your population. Using the CVS35 atlas diminishes this bias and the CVS35_inMNI152 would provide you with results in the MNI152 space, so if you need to report results in that space that could be useful.
- When using MNI152_1mm as template, I guess this is a 1 mm resolution
image (as $FREESURFER_HOME/subjects/cvs_avg35_inMNI152/mri/T1.mgz seems to be in 1 mm resolution)? Technically, is this the cvs_avg35 image that has been CVS-registered with the MNI152_1mm image? Or is this a FLIRT-type 3D registration from cvs_avg35 to MNI152_1mm without surface information (I guess extracting the surface from MNI152 would be difficult)?
It is the CVS35 average transformed and reconned in the MNI152 space. It definitely has surface information, otherwise CVS would not run.
- I run into the following error in FS stable 5.3 (source
/usr/local/freesurfer/nmr-stable53-env):
mri_cvs_check --mov $SUBJECT Using the CVS template as registration target. The following files are missing: /usr/local/freesurfer/stable5_3_0/bin/mri_cvs_register.settings.txt You need to find these files or run reconall on the data in order to run mri_cvs_register.
Is this file required, or is this a bogus error?
That file is required. I will double-check why that is not in the distribution.
- Are there parallel processing versions that could be run on the
cluster?
Yes. You can use --openmp to take adventage of it. The default is 1.
Let me know if you have any other questions. Lilla
freesurfer@nmr.mgh.harvard.edu
-
Lilla Zollei -
raij＠nmr.mgh.harvard.edu