bartorch.tools.moba

Contents

bartorch.tools.moba#

bartorch.tools.moba(kspace, TI_TE, *, r=None, L=False, P=False, F=False, T=False, G=False, D=False, bloch=False, m=None, l=None, i=None, reduction=None, j=None, u=None, C=None, s=None, B=None, b=None, d=None, N=False, f=None, p=None, J=False, M=False, O=False, g=False, positive_maps=None, not_wav_maps=None, l2_on_parameters=None, pusteps=None, ratio=None, l1val=None, temporal_damping=None, multi_gpu=None, I=None, t=None, o=None, img_dims=None, k=False, kfilter_1=False, kfilter_2=False, e=None, n=False, no_alpha_min_exp_decay=False, sobolev_a=None, sobolev_b=None, fat_spec_0=False, scale_data=None, scale_psf=None, normalize_scaling=False, seq=None, sim=None, other=None, **extra)#

Model-based nonlinear inverse reconstruction

Runs bart moba.

Parameters:

  • kspace (torch.Tensor) – Input array.

  • TI_TE (torch.Tensor) – Input array.

  • r (Regularizer | list[Regularizer] | None) – Regularization terms. (-r)

  • L (bool) – T1 mapping using model-based look-locker (-L)

  • P (bool) – T1 mapping using reparameterized (M0, R1, alpha) model-based look-locker (TR required!) (-P)

  • F (bool) – (T2 mapping using model-based Fast Spin Echo) (-F)

  • T (bool) – T2 mapping using model-based Fast Spin Echo (-T)

  • G (bool) – T2* mapping using model-based multiple gradient echo (-G)

  • D (bool) – Joint T1 and T2* mapping using model-based IR multiple gradient echo (-D)

  • bloch (bool) – Bloch model-based reconstruction (--bloch)

  • m (int | None) – Select the MGRE model from enum { WF = 0, WFR2S, WF2R2S, R2S, PHASEDIFF, …, WF_fB0, WF_R2S, T1_R2S, W_T1_F_T1_RS2 } [default: WFR2S] (-m)

  • l (int | None) – toggle l1-wavelet or l2 regularization. (-l)

  • i (int | None) – Number of Newton steps (-i)

  • reduction (float | None) – reduction factor (-R)

  • j (float | None) – Minimum regularization parameter (-j)

  • u (float | None) – ADMM rho [default: 0.01] (-u)

  • C (int | None) – inner iterations (-C)

  • s (float | None) – step size (-s)

  • B (float | None) – lower bound for relaxation (-B)

  • b (tuple[float, float] | None) – B0 field: spatial smooth level; scaling [default: 222.; 1.] (-b)

  • d (int | None) – Debug level (-d)

  • N (bool) – (normalize) (-N)

  • f (float | None) – BART’s -f. (-f)

  • p (torch.Tensor | None) – BART’s -p. (-p)

  • J (bool) – Stack frames for joint recon (-J)

  • M (bool) – Simultaneous Multi-Slice reconstruction (-M)

  • O (bool) – (Output original maps from reconstruction without post processing) (-O)

  • g (bool) – use gpu (-g)

  • positive_maps (int | tuple[int, ...] | None) – Parameter maps constrained to be positive, by index. (--positive-maps)

  • not_wav_maps (int | None) – Maps removed from wavelet denoising (counted from back!) (--not-wav-maps)

  • l2_on_parameters (int | tuple[int, ...] | None) – Parameter maps with an l2 norm, by index. (--l2-on-parameters)

  • pusteps (int | None) – Number of partial update steps for IRGNM (--pusteps)

  • ratio (float | None) – Ratio of partial updates: ratio*<updated-map> + (1-ratio)*<previous-map> (--ratio)

  • l1val (float | None) – Regularization scaling of l1 wavelet (default: 1.) (--l1val)

  • temporal_damping (float | None) – Temporal damping factor. (--temporal_damping)

  • multi_gpu (int | None) – (number of gpus to use) (--multi-gpu)

  • I (torch.Tensor | None) – File for initialization (-I)

  • t (torch.Tensor | None) – K-space trajectory (-t)

  • o (float | None) – Oversampling factor for gridding [default: 1.] (-o)

  • img_dims (tuple[int, int, int] | None) – dimensions (--img_dims)

  • k (bool) – k-space edge filter for non-Cartesian trajectories (-k)

  • kfilter_1 (bool) – k-space edge filter 1 (--kfilter-1)

  • kfilter_2 (bool) – k-space edge filter 2 (--kfilter-2)

  • e (float | None) – strength for k-space edge filter [default: 2e-3] (-e)

  • n (bool) – (disable normalization of parameter maps for thresholding) (-n)

  • no_alpha_min_exp_decay (bool) – (Use hard minimum instead of exponential decay towards alpha_min) (--no_alpha_min_exp_decay)

  • sobolev_a (float | None) – (a in 1 + a * Laplace^-b/2) (--sobolev_a)

  • sobolev_b (float | None) – (b in 1 + a * Laplace^-b/2) (--sobolev_b)

  • fat_spec_0 (bool) – select fat spectrum from ISMRM fat-water tool (--fat_spec_0)

  • scale_data (float | None) – scaling factor for data (--scale_data)

  • scale_psf (float | None) – (scaling factor for PSF) (--scale_psf)

  • normalize_scaling (bool) – (normalize scaling by data / PSF) (--normalize_scaling)

  • seq (str | None) – configure sequence parameters (--seq)

  • sim (str | None) – configure simulation parameters (--sim)

  • other (str | None) – configure other parameters (--other)

  • **extra (Any) – Further BART flags, passed through by name.

Returns:

output, sensitivities

Return type:

tuple of torch.Tensor

Contents