ReconBuffer.grid_trajectory

ReconBuffer.grid_trajectory()

Return the trajectory in grid units, laid out as bartorch.linop.NUFFT takes it.

(*placement, readout, 3): the placement axes of mask, the samples, then kx, ky and kz, each k in 1/m, as the proxy’s enrichment writes it, times the reconstructed field of view along its axis, so that an N-point image matrix spans [-N/2, N/2). Axes the acquisitions do not carry are 0. The phase-encoding axis is the last placement axis, so this is (*encoding, shots, samples, 3) against kspace as (coils, *encoding, shots, samples).

Returns:

float32; None when no acquisition carried a trajectory.

Return type:

ndarray or None

Raises:

ValueError – If the header states no field of view along an axis the trajectory carries.