bartorch.nufft#

bartorch.nufft()#

Non-uniform Fourier transform of an image to samples along traj.

Computed by FINUFFT, with a negative exponent and scaled by one over the square root of the image’s voxel count. BART’s nufft.

Parameters:

input (torch.Tensor) – Image, C order (..., z, y, x); z is one for a two-dimensional trajectory.
traj (torch.Tensor) – Trajectory (..., samples, 3) in grid units, kx, ky, kz, as bartorch.tools.traj() produces.
weights (torch.Tensor, default=None) – Diagonal in k-space applied to the samples.
basis (torch.Tensor, default=None) – Temporal subspace basis, coefficients then frames, in BART’s axis order: (coeffs, frames, 1, 1, 1, 1, 1).
**extra – Further BART nufft options, by name.

Returns:

Samples (..., *traj.shape[:-1], 1), the leading axes being those of input before (z, y, x), such as coils.

Return type:

torch.Tensor

Examples using `nufft`#