tools.reconstruct_navigator

tools.reconstruct_navigator#

bartorch.tools.reconstruct_navigator()#

Magnitude images of a navigator’s planes by density-compensated adjoint NUFFT.

Each plane is reconstructed as \(|A^H (w \odot y)|\) over its own trajectory, with \(A\) the non-uniform transform of bartorch.linop.NUFFT, and several coils are combined by root sum of squares.

Parameters:

  • kspace (torch.Tensor) – Samples (planes, samples) or (planes, coils, samples).

  • traj (torch.Tensor) – Trajectory (planes, samples, 2) in grid units of shape, kx, ky with kx along the last image axis, within \(\pm n/2\).

  • shape (tuple of int) – Matrix (y, x) of each plane.

  • density (torch.Tensor, default=None) – Density compensation weights broadcastable to (planes, samples), such as bartorch.estimate_density() returns. None weights every sample by one.

Returns:

Real float32 magnitudes (planes, *shape), on the device of kspace, for NavigatorMotionTracker.track().

Return type:

torch.Tensor

Raises:

ValueError – If the trajectory does not match the samples, or leaves the grid.

Examples using reconstruct_navigator#

Rigid head motion from navigators

Rigid head motion from navigators