.. _autofocus_top:

Coregister with AutoFocus
=========================

.. only:: html

   .. sidebar:: Tutorials & Sample Data

      Watch on YouTube: |br|
      `Tutorial 13: 3D Image Coregistration Using Mutual Information <https://youtu.be/b_kRwYyaVac>`_

      :download:`Tut13data.zip (6.5MB) <https://www.firevoxel.org/im/Tut13data.zip>`

.. contents::
   :depth: 1
   :local:
   :backlinks: top

.. _autofocus_intro:

About Coregistration with AutoFocus
-----------------------------------

**AutoFocus** is a scheme that iteratively optimizes a voxel-based similarity
measure between two images, such as mutual information. Coregistration with AutoFocus
is a powerful and versatile coregistration method that can be used for coregistering
images from different imaging sessions and different modalities.
For images acquired in the same session, see the basic method for
:ref:`coregistration using DICOM tags <register_tags_top>`.

This section describes AutoFocus coregistration for 3D images.
FireVoxel also offers **AutoFocus with motion correction** for 4D coregistration.

.. _autofocus_basics:

Basics of AutoFocus
-------------------

During coregistration, the **source image** is transformed and superimposed onto
a fixed image called the **target image**, which remains unchanged.

Coregistration with AutoFocus may use a **target ROI** enclosing the organ or tissue
of interest to restrict coregistration and speed up processing. Coregistration *can be*
performed without the target ROI, but it may take much longer than with the ROI,
because coregistration is a computationally intensive task.

The target ROI can be created on the **target image** manually (using
the :ref:`Paintbrush <paintbrush_create_roi>` tool) or automatically, using
:ref:`ROI <roi_top>` operations or segmentation tools, such as :ref:`EdgeWave <edgewave_top>`.

The coregistration algorithm searches for a transformation that best matches
the source and the target volumes. This group of commands offers a choice
of transformations ranging from simple translations to affine transform.
The matching of the volumes is based on optimizing a **similarity measure**
(with a selection for different scenarios).

The transformation is computed in two stages, **AutoFocus** and **Fine-tuning**:

1. **AutoFocus.** The algorithm constructs a variety of transformations
with combinations of parameters that span a multidimensional grid.
The transformations include translation, scaling, rotation, and shear.
The transformations are ranked by how well they match the two volumes based
on the similarity measure. A user-selected number of the best transformations
is retained for the second, fine-tuning stage.

2. **Fine-tuning.** The algorithm performs iterative adjustment
of the best transformation parameters until it finds a local optimum
of the similarity measure. Finally, the transformed source image is interpolated
and saved as a new layer in the target image window.

.. _autofocus_mutual_info:

Using Coregistration with AutoFocus
-----------------------------------

Here we will describe coregistration with AutoFocus using
**Register** > **Mutual Information with AutoFocus**. Other variants
of AutoFocus commands are applied similarly.

1. :ref:`Open <open_top>` the target window
in another document window.

2. Open the target window
in another document window.

3. Define a rough target ROI around the organ or tissue of interest.
You may use manual or automatic segmentation tools (:numref:`fig_coreg_mi_setup`).

**Manual ROI.** Use **Layer Control** > :ref:`New ROI 3D` to create a new ROI layer.
Use the :ref:`Paintbrush <paintbrush_create_roi>` tool (Ctrl+Left mouse) to draw a rough
contour around the organ of interest. Define these contours on every few slices
(e.g., on every 5th slice). Next, use **ROI** > **Morphology** >
:ref:`Fill 2D Contours and Morph Convex <roi_fill_2d_contours>` to fill the contours
and extend the ROI across slices. The resulting ROI should fully enclose the organ
or tissue of interest.

   .. _fig_coreg_mi_setup:
   .. figure:: ../images/coreg_mi_af_setup_800.png
      :alt: Target image with target ROI and source image
      :scale: 70%
      :align: center
      :figclass: align-center

      Target image with target ROI (MR, right) and source image (CT, left).

4. Select the target window as the active window and select the base image as
the active layer. Select **Register** > **Mutual Information with AutoFocus**.

5. If the active layer is *an ROI layer*, a warning will be shown alerting the user
to this fact (**Target is ROI. Proceed?**).

6. If the active layer is *the base image* (acquired image), a dialog window
will open to adjust parameters (:ref:`3D Registration with AutoFocus <autofocus_3d>`).

7. Check **Use Target ROI**
box (checked by default if a *visible ROI* is present).
If no *visible ROI* is present, this box is grayed out.

8. Select suggested **AutoFocus** and **Finetune** parameter
(:numref:`fig_autofocus_dialog`). It is recommended to start with
**AutoFocus: Translation** and **Rotation** only and **Finetune: Transform > Rigid**.

9. Click **OK** to start processing. Registration will commence.
Its progress will be indicated in the bottom right corner of the software window.

10.	When registration is completed, a new layer with coregistered image will be created
in the target window (:numref:`fig_coreg_mi_result`).

   .. _fig_coreg_mi_result:
   .. figure:: ../images/coreg_mi_af_result_800.png
      :alt: Coregistration with AutoFocus results
      :scale: 70%
      :align: center
      :figclass: align-center

      Coregistration with AutoFocus results.

11. Check registration accuracy using :ref:`Layer Control <layer_control_panel>`
> :ref:`Alpha <layer_transparency>` slider to adjust the transparency
of the coregistered layer.

12. If registration accuracy is unacceptable, repeat steps 5-11 and adjust
the coregistration parameters. As the first step, increase **Power** before
adjusting other parameters.

.. _autofocus_3d:

3D Registration with AutoFocus Dialog
-------------------------------------

If suitable source and target images are present, **Register** >
**Mutual Information with AutoFocus** command opens a dialog panel
(**3D Registration with AutoFocus**, :numref:`fig_autofocus_dialog`).
The components of this dialog are described below.

If one of the volumes is missing, an error message is shown
(**No suitable source volume is found**).

   .. _fig_autofocus_dialog:
   .. figure:: ../images/coreg_mi_af_dialog.png
      :alt: 3D Registration with AutoFocus dialog
      :align: center
      :figclass: align-center

      3D Registration with AutoFocus dialog.

.. _autofocus_load_initial:

Load Initial Transformation
~~~~~~~~~~~~~~~~~~~~~~~~~~~

Loads transformation information from a previously saved
:ref:`Volume Transform File <tx_vtf>` (\*.VTF).
Enter a path to the .VTF into the text box or click **Browse** to open
browse-for-file dialog to select a previously saved transform file.

**Dicom tags** -- :red:`ADD DETAIL`.

Save Final Transformation
~~~~~~~~~~~~~~~~~~~~~~~~~

Save the final transformation as \*.VTF.
Enter the path to the file or click **Browse** to open browse-for-file dialog
to navigate to the destination directory and enter a file name.

.. _autofocus_roi:

ROI (option)
~~~~~~~~~~~~

**Use Target ROI** (checkbox) -- If checked, the target ROI is used for
coregistration. If a *visible ROI* is present in the target window,
by default its name will be displayed here.
If the ROI is absent or invisible, this option is grayed out.

**Inflate/units** -- Text box and drop-down menu to select the *grow distance*
(in voxels :red:`OPTION: MILLIMETERS?`) by which the target ROI should be
inflated by the :ref:`Grow <roi_grow>` command.

.. _autofocus_measure:

Measure (options)
~~~~~~~~~~~~~~~~~

This block contains a dropdown menu with a selection of similarity measures
and related settings. The **similarity measures** range from simple
to complex and powerful:

    - Signal Difference (simple)

    - Cross Correlation

    - Image Ratio Uniformity

    - Mutual Info

    - Mutual Info Normalized

    - URAL

    - URALTAU (advanced)

**Related settings include:**

    - MI bin number -- Number of bins in the mutual information method

    - Source Noise -- Noise level in the source image (default: measured automatically)

    - Target Noise -- Noise level in the target image (default: measured automatically).

**URAL Settings** -- Opens dialog
(:ref:`Multiscale Texture Gradient <volume_texture_gradient_multiscale>`,
:numref:`fig_multiscale_grad`) to adjust :ref:`texture edge detector <texture_edge_top>`
parameters for URAL and URALTAU methods.

.. _autofocus_options:

AutoFocus (options)
~~~~~~~~~~~~~~~~~~~

Subsample [1, 8] -- Default 3

Translation max {X, Y, Z} -- Default: {30, 30, 30}

Scale Deformation matrix -- Default: {0, 0, 0}, Grid: 1

Uniform scale (checkbox) -- Default -- unchecked

Rotation angle max (deg) -- Default: {0, 0, 0}, Grid: 1

Shear Magnitude max [0, 10] -- Default: 0, Grid: 1

.. _autofocus_finetune:

Finetune (options)
~~~~~~~~~~~~~~~~~~

Power [0, 1000] -- Default: 20. Suggested starting value: 2-3

Multipass (checkbox) -- Default: unchecked

Transform (dropdown menu) -- Dropdown menu with a selection of transform types:
Translation, Rigid (default), Affine, Quadratic, Polynomial Multipass.

.. _autofocus_output:

Output (options)
~~~~~~~~~~~~~~~~

Interpolation (dropdown menu) -- Nearest neighbor, Tri-linear (default),
Wsinc2, Wsinc3, Wsinc4

Reslice Target to Source (checkbox) -- Default: unchecked.

.. _autofocus_signal_diff:

Signal Difference with AutoFocus
--------------------------------

Shortcut for coregistration with Similarity Measure: Signal Difference

.. _autofocus_cross_correlation:

Cross Correlation with AutoFocus
--------------------------------

Shortcut for coregistration with Similarity Measure: Cross Correlation

.. _autofocus_ural:

URAL with AutoFocus
-------------------

Shortcut for coregistration with Similarity Measure: URAL

.. _autofocus_slice:

Slice-by-slice with AutoFocus
-----------------------------

Requires source and target images of the same dimensions.


|top|