Registration algorithm: ``sct_register_to_template``
####################################################

.. figure:: https://raw.githubusercontent.com/spinalcordtoolbox/doc-figures/master/template-registration/thin-plate-straightening.png
   :align: right
   :figwidth: 300px

   Spinal cord straighting using thin-plate spline interpolation.

SCT provides the ``sct_register_to_template`` command for template registration. Here are the steps for the algorithm within this command:

1. **Straightening:** The straightening works by finding, for each point along the spinal cord, the mathematical transformation to go from a curved centerline to a straight centerline.

   * The straightening algorithm computes the orthogonal plane at each point along the centerline, then constructs a straight space in the output using thin-plate spline interpolation. This allows the inner geometry of the cord to be preserved.
   * The straightening algorithm outputs a forward and a backward warping field (ITK-compatible), which can be concatenated with subsequent transformations, as will be seen later.

2. **Vertebrae-matching transformation**: Once straightened, the next step involves a transformation to match the vertebral levels of the subject to that of the template. If 2 labels are provided, this transformation will be affine; if 3+ labels are provided, this transformation will be non-affine. (Note: This step focuses only on matching the coordinates of the labels, and does not consider the shape of the spinal cord, which is handled by the next step.)
3. **Shape-matching transformation**: A multi-step nonrigid deformation is estimated to match the subject’s cord shape to the template. By default, two steps are used: the first handles large deformations, while the second applies fine adjustments.

.. warning::

   SCT provides many additional nonrigid deformation algorithms beyond the default configuration. You can visit the :ref:`customizing-registration-section` page to learn how to optimize the registration procedure for your particular contrast, resolution, and spinal cord geometry.