apps.nlinv_pics

Contents

apps.nlinv_pics#

bartorch.apps.nlinv_pics()#

Wavelet-regularized SENSE reconstruction with coil sensitivities fitted to the data.

The sensitivities are those of nlinv_maps(). The image is the pics() minimizer of |P F S x - y|^2 + lambda |W x|_1 (bart pics -R W). For Cartesian data, an axis acquired on one side only is then completed by partial_fourier(). Unsampled k-space is zero: the sampling mask is the support of kspace.

Parameters:

  • kspace (torch.Tensor) – Complex k-space. Cartesian: (coils, [z,] y, x) on the reconstruction grid. Non-Cartesian: (coils, shots, samples).

  • traj (torch.Tensor, default=None) – Trajectory (shots, samples, 3) in units of the image grid; None for Cartesian data.

  • wavelet (float, default=0.005) – lambda, relative to the data scaling pics estimates.

  • iterations (int, default=30) – Iterations of the pics solve.

  • size (int, default=24) – Cartesian: lines of every encoded axis, around the centre, the sensitivities are fitted to.

  • radius (float, default=12.0) – Non-Cartesian: distance from the k-space centre, in grid units, within which samples are fitted.

  • phase_pole (int, default=None) – Gauss-Newton step of nlinv after which phase poles of the sensitivities are moved into the image; 0 after every step, None never (nlinv_maps()).

Returns:

Complex image in the units of kspace: ([z,] y, x) as kspace for Cartesian data, (y, x) on the image grid of the trajectory otherwise.

Return type:

torch.Tensor

Contents