mr2.data.traj_calculators.KTrajectoryCartesian

class mr2.data.traj_calculators.KTrajectoryCartesian[source]

Bases: KTrajectoryCalculator

Cartesian trajectory.

__init__()

classmethod fullysampled(encoding_matrix: SpatialDimension[int]) → KTrajectory[source]

Generate fully sampled Cartesian trajectory.

Parameters:

encoding_matrix (SpatialDimension[int]) – Encoded K-space size.

Returns:

Cartesian trajectory.

classmethod gaussian_variable_density(encoding_matrix: SpatialDimension[int] | int, acceleration: float = 2.0, n_center: int = 10, fwhm_ratio: float = 1.0, n_other: Sequence[int] = (1,), seed: int | None = None) → KTrajectory[source]

Generate k-space Gaussian weighted variable density sampling.

Parameters:

• encoding_matrix (Union[SpatialDimension[int], int]) – Encoded K-space size, must have encoding_matrix.z=1. If a single integer, a square k-space is considered.

• acceleration (float, default: 2.0) – Acceleration factor (undersampling rate).

• n_center (int, default: 10) – Number of fully-sampled center lines to always include.

• fwhm_ratio (float, default: 1.0) – Full-width at half-maximum of the Gaussian relative to encoding_matrix.y. Larger values approach uniform sampling. Set to infinity for uniform sampling.

• n_other (Sequence[int], default: (1,)) – Batch size(s). The trajectory is different for each batch sample.

• seed (int | None, default: None) – Random seed for reproducibility.

Returns:

Cartesian trajectory.

Raises:

• ValueError – If n_center exceeds the total number of lines to keep given the acceleration.

• NotImplementedError – If called with a 3D encoding matrix.

classmethod uniform_undersampling(encoding_matrix: SpatialDimension[int] | int, acceleration: float = 2.0, n_center: int = 10) → KTrajectory[source]

Generate deterministic uniformly undersampled Cartesian sampling.

Every acceleration-th line in k1 is sampled, and center k-space lines are always sampled via n_center.

Parameters:

• encoding_matrix (Union[SpatialDimension[int], int]) – Encoded K-space size, must have encoding_matrix.z=1. If a single integer, a square k-space is considered.

• acceleration (float, default: 2.0) – Uniform undersampling factor. Must be an integer value >= 1. For example, acceleration=4 samples every 4th line.

• n_center (int, default: 10) –

  Number of fully-sampled center lines to always include.

  Note: if n_center > 0, the effective acceleration is lower than acceleration because ACS lines are added on top of regular samples.

Returns:

Cartesian trajectory.

Raises:

• ValueError – If acceleration is invalid or n_center is outside [0, n_k1].

• NotImplementedError – If called with a 3D encoding matrix.

__call__(*, n_k0: int, k0_center: int | Tensor, k1_idx: Tensor, k1_center: int | Tensor, k2_idx: Tensor, k2_center: int | Tensor, reversed_readout_mask: Tensor | None = None, **_) → KTrajectory[source]

Calculate Cartesian trajectory for given KHeader.

Parameters:

• n_k0 (int) – number of samples in k0

• k0_center (int | Tensor) – position of k-space center in k0

• k1_idx (Tensor) – indices of k1

• k1_center (int | Tensor) – position of k-space center in k1

• k2_idx (Tensor) – indices of k2

• k2_center (int | Tensor) – position of k-space center in k2

• reversed_readout_mask (Tensor | None, default: None) – boolean tensor indicating reversed readout

Returns:

Cartesian trajectory for given KHeader

static gaussian_variable_density_nd(encoding_matrix: SpatialDimension[int], acceleration: SpatialDimension[float], n_center: SpatialDimension[int] | int = 10, fwhm_ratio: SpatialDimension[float] | float = 1.0, n_other: Sequence[int] = (1,), seed: int | None = None) → KTrajectory[source]

Generate Gaussian weighted variable density Cartesian undersampling.

Undersampling is applied independently in each dimension where acceleration > 1.

Parameters:

• encoding_matrix (SpatialDimension[int]) – Encoded K-space size.

• acceleration (SpatialDimension[float]) – Per-axis acceleration factors. Dimensions with acceleration <= 1 are fully sampled.

• n_center (Union[SpatialDimension[int], int], default: 10) – Number of fully-sampled center lines to always include per axis. If given as an integer, the same value is used for all axes.

• fwhm_ratio (Union[SpatialDimension[float], float], default: 1.0) – Per-axis full-width at half-maximum of the Gaussian relative to the encoded matrix size. Larger values approach uniform sampling. If given as a scalar, the same value is used for all axes.

• n_other (Sequence[int], default: (1,)) – Batch size(s). The trajectory is different for each batch sample.

• seed (int | None, default: None) – Random seed for reproducibility.

Returns:

Cartesian trajectory.

Raises:

ValueError – If any acceleration is non-positive, if any n_center is outside the corresponding encoding range, or if a requested center region exceeds the number of retained samples along an undersampled axis.

__eq__(value, /)

Return self==value.

__new__(**kwargs)