• autoreg-sess

Index

Name

autoreg-sess

Synopsis

autoreg-sess -s <subjectname> -d <subjectdirectory>

Arguments

Positional Arguments

none

Required Flagged Arguments

Optional Flagged Arguments

Outputs

Description

The functional volumes for each subject need to be registered with the anatomical volume in the Anatomical Database. This is difficult to do because the two volumes have different resolutions and different contrasts and may have been collected on different sessions. To mitigate this problem, the registration will be accomplished in two stages: Automatic and Manual. The automatic is used to get the registration close, and the manual is used to fine-tune the registration.

Automatic Stage

The automatic registration can also be broken into two conceptual steps, though both are accomplished with one command autoreg-sess. The first step is the Header Registration in which the functional volumes are registered with the anatomicals that were collected during the same session. No MRI data is used in the header registration. It relies on the assumption that the head has not moved so that registration can be accomplished simply by registering the fields-of-view. In the second step, the same-session anatomical is registered with the Database anatomicals. This is accomplished by translating/rotating the volumes based on a voxel-by-voxel comparison. This can be done because both volumes have T1 contrast. The automatic registration is executed by running autoreg-sess from inside the Study Directory with the –sf and –df options. This will create a file called register.dat in each functional subdirectory. This represents the registration between the functional volume and the Database anatomical volume.

Manual Stage

Variation 1

Because of movement and EPI distortion, the automatic stage can only get the registration close. The manual stage allows the user to check and fine-tune the automatic registration. To run the manual registration, cd to the Study Directory and run tkregister-sess with the –sf and –df options. This will read the register.dat created by the automatic registation and bring up two windows: an image window and a control window. The image window will initially have a coronal view of the anatomical in the Database. Hitting the COMPARE button will switch views to the functional volume. When the volumes are properly registered, the anatomical features in the functional slice should perfectly overlay those in the anatomical slice. Hitting the SAGITTAL button will change the view to sagittal. Hitting the HORIZONTAL button will change the view to axial. The slider under each view button can be used to change the slice viewed. The current slice can also be changed by clicking in the image window (moving the red cursor). With these operations, the user can navigate through the volume and check the registration. If the registration is off, then the user has three options: translation, rotation, and scaling, each of which has it’s own frame inside the control window. Within the Translation frame, there are two sliders. The vertical slider translates the functional brain in the vertical direction, where “vertical” is with respect to the current view in the image window. The horizontal slider translates the functional brain horizontally. The Rotation frame has only one slider. Moving the slider will rotate the functional brain around the current cursor location about the axis perpendicular to the current image view. The Scale frame has two sliders. The vertical slider will stretch/compress the functional brain in the current vertical direction. The horizontal slider does the same for the current horizontal direction. When finished, hit the SAVE REG button (it may prompt you to overwrite the current registration; if so, do so). To quit, hit the button on the upper left hand corner to bring down a menu and choose the Close option. This will update the register.dat. Note that re-running autoreg-sess will cause the results from the manual registration to be overwritten. Also note that any changes in registration will require that all down-stream operations (eg, resampling, inter-subject averaging) be rerun.

Variation 2

Manual registration of t2conv before registration of functionals. One of the reasons that it is difficult to register the functionals is because they are typically very low resolution. One can, however, first manually register the T2 Conventionals which are co-planar to the functionals but have 2-4 times the in-plane resolution. To do this, call tkregister-sess with “-rsd t2conv”. This will create a register.dat in the t2conv directory, and the registration matrix inside register.dat will be copied to the register.dat in the bold directory. Do not attempt to perfectly register the t2conv volume, just use it to get the registration close. After registering with t2conv, re-run tkregister-sess without the “-rsd t2conv” to perform the final functional registrataion.

Tips for Manual Registration

Manual registration can be a tedious process. It is recommended that the user first translate, then rotate, and, as a last resort, scale. Keep in mind that a registration improvement in one part of the volume can degrade that in another part. It is best to make small adjustments at several locations around the brain, slowly converging on the optimum registration

Examples

Example 1: Variation 1

autoreg-sess -s <subjectname> -d <subject_dir>

tkregister-sess -s <subjectname> -d . (here you are registering anatomicals to functionals)

Example 2: Variation 2

autoreg-sess -s <subjectname> -d . <subject_dir>

tkregister-sess -s <subjectname> -d . -rsd t2conv (here you are registering the anatomicals to the t2conv that was collected in the same session as the functionals. This will assist in aligning anatomicals to functionals)

tkregister-sess -s <subjectname> -d . (here you are registering anatomicals to functionals)

Bugs

None

See Also

tkregister-sess, tkregister2

Links

FreeSurfer, FsFast

Methods Description

References

References/Lastname###

Reporting Bugs

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

Author/s

DougGreve, PhD