Module: Image Registration Wizard ()

Description:

This module computes and applies a transformation to align two images:

• Step 1 - Display
  Optional operation for displaying data sets with Ortho Views.
  First select the model and the reference data sets. The model is the image to be aligned. Click on "Apply" to display Ortho Views.
  Click on "Skip" if no new displaying module is required. Additionally, choose "Hide module" to hide all modules used by this script.
• Step 2 - Model simplification
  Subsample and extract subvolume from model if needed.
  In this step, the model is simplified and/or reduced before alignment step. In case of model and reference representing the same volume :
• Select the whole volume by clicking on Box Max buttons.
• Select Subsample option if the data set is heavy.
You can also click on "Skip" if no simplification is required.
In case of overlapped volumes (model and reference have a limited common region) :
• Select the region of interest using the blue dragger in the viewer, or by typing box's lower corner indexes and size in Box Min and Box Size ports.
• Select Subsample option if the data set is heavy.
See Extract Subvolumes for more information.
• Step 3 - Reference simplification
  Subsample and extract subvolume from reference if needed.
  See the previous step description.
• Step 4 - Registration
  Alignment of the simplified model with the simplified reference.
  Set the metric, the transformation type and the pre-alignment method and click on the "Apply" button.
• The Metric depends on data sets. In most of cases, Correlation is enough. If images are from different modality, Mutual Information metric are suitable. Otherwise, Euclidean is suitable.
• Histogram Range Reference and Model are parameters for Mutual Information metrics.
• The Transformation Type depends on initial state of the two volumes. If the model and the reference have the same scale : choose Rigid transformation. If the volumes have a different but uniform scale, choose Iso-scale option. Aniso-scale option allows different scaling on different axes and the Shear options allows shearing.
• If images are from same modality, use Align Centers pre-alignment method. If they have unambiguous principal axis, Align Principal Axes is suitable. If they are already approximately aligned, select None option.
At the end of this step, if the alignment is not correct : try out different metrics or pre-alignment methods.
See Register Images for the choice of parameters.

Connections:

Data [required]
Image data to be transformed.

Reference [required]
Reference image.

Ports:

Info

This port displays information about the current action.

Instruction

This port displays instruction about the current action.

Action

This port contains the 3 buttons "Back", "Skip" and "Apply" triggering the actions to perform.

Hide module

[Step 1] If this option is enabled, all new modules will be hiden during the computation.

Box min

[Step 2 and 3] The lattice index of the lower corner of the subblock. The subblock corresponds to the volume's region to consider for registration.

Box size

[Step 2 and 3] The size of the subblock to be loaded. The size is in lattice indices in the base resolution. The subblock corresponds to the volume's region to consider for registration.

Box max

[Step 2 and 3] Convenience buttons to use the maximum box size for the width, height and depth, respectively. If Max width, Max height and Max depth buttons are hit, the whole volume is selected as region of interest for alignment.

Subsample

[Step 2 and 3] This port enables subsampling.

Subsample rate

[Step 2 and 3] If subsample is enabled, this port defines the number of voxels to be averaged in each dimension.

Metric

[Step 4] This port selects the similarity measure to be applied for registration. See Register Images documentation.

Histogram range reference

[Step 4] This port defines Mutual Information metrics parameters. The range of gray values for the histogram of the reference data set is defined at this port. See Register Images documentation.

Histogram range model

[Step 4] This port defines Mutual Information metrics parameters. The range of gray values for the histogram of the model data set is defined at this port. See Register Images documentation.

Transformation

[Step 4] This port defines the number of transformation parameters to be optimized during the alignment process. Six parameters are optimized for Rigid (3 translations and 3 rotations), 7 for Iso-scale (add 1 parameter of scaling), 9 for Aniso-scale (3 translations, rotations and scalings) and 12 for Shear. See Register Images documentation.

Pre-alignment

[Step 4] This port defines the pre-alignment method. If Align center is selected, the centers of gravity of both data sets will be aligned before registration. If Align principal axes is selected, the centers of gravity and the moments of inertia of both data sets are computed and aligned, by testing all 24 possible alignments : the best alignment is selected using the similarity measure defined by Metric. If None is selected, no pre-alignment will be done. See Register Images.