mr2.algorithms.image_registration.affine_registration

mr2.algorithms.image_registration.affine_registration(fixed: Tensor, moving: Tensor, *, weight: Tensor | None = None, downsampling_factor: int = 1, window_size: int = 9, max_iterations: int = 50, regularization_weight: float = 1e-4, initial_affine: Tensor | None = None) → GridSamplingOp[source]

Run one affine registration level.

Parameters:

• fixed (Tensor) – Fixed image.

• moving (Tensor) – Moving image.

• weight (Tensor | None, default: None) – Weight/mask tensor. If None, no weighting is applied.

• downsampling_factor (int, default: 1) – Downsampling factor.

• window_size (int, default: 9) – Window size for NCC similarity measure.

• max_iterations (int, default: 50) – Maximum number of iterations of the LBFGS optimizer.

• regularization_weight (float, default: 1e-4) – Regularization weight for L2 regularization of the affine matrix.

• initial_affine (Tensor | None, default: None) – Initial affine matrix. If None, the identity matrix is used.