Differences between revisions 5 and 6
Deletions are marked like this. Additions are marked like this.
Line 91: Line 91:
************

You will not be able to use fsaverage as a template as it does not have a
corresponding surface and volumetric atlas. We are in the process of
building and releasing a CVS atlas, but it is not ready yet, so for now
the best thing to do would be to register all your subjects to a randomly
selected one.

Index

Name

mri_cvs_register - registers MRI scans of two subjects (or a subject and the CVS atlas) using the combined volume and surface-based registration method, CVS. Make sure that SUBJECTS_DIR is properly set before calling this script!

Synopsis

mri_cvs_register--mov arg1 --template arg2

Arguments

Required Flagged Arguments

--template subjid

subjid for subject to be moved / registered

Subject id of the subject (as found in SUBJECTS_DIR) whose scan is to be moved in registration with that of the template.

--mov subjid

subjid for subject to be kept fixed (template)

Subject id of the subject (as found in SUBJECTS_DIR) whose scan is to be used and the template / target for the registration.

Optional Flagged Arguments

--outdir directory

output directory where all the results are written

Output directory where the resulting morphs should be written. NOTE, the morphs require quite a lot of space so make sure that there is enough disk space available before your analysis. The default location for the output is: SUBJECTS_DIR/mov/cvs

--noaseg

do not use aseg volumes in the registration pipeline (default is 0)

--nocleanup

do not delete temporary files (default is 0)

--nolog

do not produce a log file (default is 0)

--step1

only do step 1 (spherical registration)

--step2

only do step 2 (elastic registration)

--step3

only do step 3 (volumetric registration)

--version

print version and exit

--help

print help and exit

Outputs

See below.

Description

The CVS registration stream consists of 3 steps: (1) surface-based registration, (2) elastic registration and (3) volumetric regsitration. By default, in the current implementation of the script, step (3) consists of two steps: (i) registration using aseg volumes and (ii) registration using intensity volumes. This combination of the non-linear volumetric registrations proved to be very robust and accurate, butthe (i) step could be eliminated if quicker execution time is required by using the "--noaseg" flag.

Relevant files that you should have in your directory (most recent on the top and the oldest in the bottom):

(a) Full CVS (with both (i) and (ii)) nlalign-afteraseg-norm.m3z : the morph resulting from (ii) of step (3) (intensity-based non-linear registration); it is NOT combined with the elastic registration morph nlalign-afteraseg-norm.mgz : the morphed norm.mgz file; it is the final result of CVS and contains contributions from ALL registration steps (combination of the elatic morph and the two m3z files)

nlalign-aseg.m3z : the morph resulting from (i) of step (3) (aseg-based non-linear registration); it is NOT combined with the elastic registration morph nlalign-aseg.mgz : the morphed aseg.mgz file; it is the result of CVS up to (i) of step (3) and contains contributions from registration steps of the elatic morph and the above m3z file

el_reg_toTEMPLATE_aseg.mgz : the elastic morph applied to aseg.mgz el_reg_toTEMPLATE.tm3d : the elastic morph that combines correspondences recovered in steps (1) and (2) el_reg_toTEMPLATE.mgz : the elastic morph applied to norm.mgz

(b) shorter version of CVS (using only (ii) in step (3) of the algorithm):

nlalign-norm.m3z : the morph resulting from step(3) (intensity-based non-linear registration); it is NOT combined with the elastic registration morph nlalign-norm.mgz : the morphed norm.mgz file; it is the final result of CVS and contains contributions from ALL registration steps (combination of the elatic morph and the above m3z file)

el_reg_toTEMPLATE.tm3d : the elastic morph that combines correspondences recovered in steps (1) and (2) el_reg_toTEMPLATE.mgz : the elastic morph applied to norm.mgz

If you ran the mri_cvs_registration script with the "--nocleanup" option then you will see two more files called combined*tm3d. These files take up a lot of memory so by default are deleted. They can be recreated in case the cvs morph needs to be applied to another volume at a later time. The morphs left behind when using the "--nocleanup" option are:

(a) el_reg_toTEMPLATE.tm3d : the elastic morph that combines correspondences recovered in steps (1) and (2) combined_elreg_aseg.tm3d : the morph that combines correspondences recovered in steps (1), (2) and (3) (i) combined_elreg_afteraseg-norm.tm3d : the morph that combines correspondences recovered in steps (1), (2), (3) (i) and (3) (ii) (b) el_reg_toTEMPLATE.tm3d : the elastic morph that combines correspondences recovered in steps (1) and (2) combined_elreg_norm.tm3d : the morph that combines correspondences recovered in steps (1), (2) and (3) (ii)

***********

If you want to reconstruct the resulting warp (as it was deleted to save space after the processing) and apply it to a new volume, you would need to call two functions. If the warp combined*tm3d does not exist run both (1) an d(2) , if it already exists, just run (2).

(1) createMorph --out fullCVSmorph.tm3d --template template.mgz \

  • --subject movingvol.mgz --in gcam nlalign-afteraseg-norm.m3z \ morph el_reg_toTEMPLATE.tm3d

(2) applyMorph --template template.mgz --transform fullCVSmorph.tm3d \ vol movingvol.mgz outputvol.mgz linear

************

Note, the longest processing step in the CVS pipeline at the moment is the volumetric registration which might take about 15 hrs. The rest takes about 2 hours.

************

You will not be able to use fsaverage as a template as it does not have a corresponding surface and volumetric atlas. We are in the process of building and releasing a CVS atlas, but it is not ready yet, so for now the best thing to do would be to register all your subjects to a randomly selected one.

Bugs

None

See Also

mris_register, mri_nl_align, mris_resample, mri_cvs_check, mri_cvs_data_copy

Links

FreeSurfer

References

G.M. Postelnicu*, L. Zöllei*, B. Fischl: "Combined Volumetric and Surface Registration", IEEE Transactions on Medical Imaging (TMI), Vol 28 (4), April 2009, p. 508-522

L. Zöllei, A. Stevens, K. Huber, S. Kakunoori, B. Fischl: "Improved Tractography Alignment Using Combined Volumetric and Surface Registration", NeuroImage 51 (2010), 206-213

Reporting Bugs

Report bugs to <analysis-bugs@nmr.mgh.harvard.edu>

Author/s

LillaZollei

mri_cvs_register (last edited 2018-02-05 12:34:44 by MorganFogarty)