Tomobar Recon Cpu¶

Description¶

A CPU Plugin to reconstruct full-field tomographic projection data using state-of-the-art regularised iterative algorithms from the ToMoBAR package. ToMoBAR includes FISTA and ADMM iterative methods and depends on the ASTRA toolbox and the CCPi RGL toolkit.

Parameters

in_datasets:
    visibility: datasets
    dtype: "[list[],list[str]]"
    description:
        summary: A list of the dataset(s) to process.
        verbose: A list of strings, where each string gives the name of a dataset that was either specified by a loader plugin or created as output to a previous plugin.  The length of the list is the number of input datasets requested by the plugin.  If there is only one dataset and the list is left empty it will default to that dataset.
    default: "[]"

out_datasets:
    visibility: datasets
    dtype: "[list[],list[str]]"
    description:
        summary: A list of the dataset(s) to create.
        verbose: A list of strings, where each string is a name to be assigned to a dataset output by the plugin. If there is only one input dataset and one output dataset and the list is left empty, the output will take the name of the input dataset. The length of the list is the number of output datasets created by the plugin.
    default: "[]"

centre_of_rotation:
    visibility: basic
    dtype: "[float, str, dict{int:float}]"
    description: Centre of rotation to use for the reconstruction.
    default: "0.0"
    example: It could be a fixed value, a dictionary of (sinogram number, value) pairs for a polynomial fit of degree 1, or a dataset name.

init_vol:
    visibility: intermediate
    dtype: "[None, str]"
    description: Dataset to use as volume initialiser (does not currently work with preview)
    default: None
    example: "Type the name of the initialised dataset e.g. ['tomo']"

log:
    visibility: intermediate
    dtype: bool
    description:
        summary: Option to take the log of the data before reconstruction.
        verbose: Should be set to false if you use PaganinFilter
    default: "True"
    example: Set to True to take the log of the data before reconstruction.

preview:
    visibility: intermediate
    dtype: preview
    description: A slice list of required frames.
    default: "[]"

force_zero:
    visibility: intermediate
    dtype: "[list[float,float],list[None,None]]"
    description: Set any values in the reconstructed image outside of this range to zero.
    default: "['None', 'None']"
    example: "[0, 1]"

ratio:
    visibility: intermediate
    dtype: "[float, list[float, float]]"
    description: Ratio of the masks diameter in pixels to the smallest edge size along given axis. If a list of two floats is given, the second value is used to fill up the area outside the mask.
    default: "0.95"

log_func:
    visibility: advanced
    dtype: str
    description: Override the default log function with a numpy statement
    default: np.nan_to_num(-np.log(sino))

vol_shape:
    visibility: intermediate
    dtype: "[str, int]"
    description:
        summary: Override the size of the reconstruction volume with an integer value.
        verbose: When fixed, you get the dimension of the horizontal detector or you can specify any reconstruction size you like with an integer.
    default: fixed

data_fidelity:
    visibility: advanced
    dtype: str
    description: Data fidelity, choose LS, PWLS, SWLS or KL.
    default: LS

data_Huber_thresh:
    visibility: advanced
    dtype: "[None,float]"
    description: Threshold parameter for __Huber__ data fidelity.
    default: None

data_beta_SWLS:
    visibility: advanced
    dtype: float
    description: A parameter for stripe-weighted model.
    default: "0.1"

data_full_ring_GH:
    visibility: advanced
    dtype: "[None,str]"
    description: Regularisation variable for full constant ring removal (GH model).
    default: None

data_full_ring_accelerator_GH:
    visibility: advanced
    dtype: float
    description: Acceleration constant for GH ring removal. (use with care)
    default: "10.0"

algorithm_iterations:
    visibility: basic
    dtype: int
    description:
        summary: Number of outer iterations for FISTA (default)or ADMM methods.
        verbose: Less than 10 iterations for the iterative method (FISTA) can deliver a blurry reconstruction. The suggested value is 15 iterations, however the algorithm can stop prematurely based on the tolerance value.
    default: "15"

algorithm_verbose:
    visibility: advanced
    dtype: str
    description: Print iterations number and other messages (off by default).
    options: "['on', 'off']"
    default: off

algorithm_mask:
    visibility: advanced
    dtype: float
    description: set to 1.0 to enable a circular mask diameter or < 1.0 to shrink the mask.
    default: "1.0"

algorithm_ordersubsets:
    visibility: advanced
    dtype: int
    description: The number of ordered-subsets to accelerate reconstruction.
    default: "6"

algorithm_nonnegativity:
    visibility: advanced
    dtype: str
    options: "['ENABLE', 'DISABLE']"
    description:
        summary: ENABLE or DISABLE nonnegativity constraint.
    default: ENABLE

regularisation_method:
    visibility: intermediate
    dtype: str
    options: "['ROF_TV', 'FGP_TV', 'PD_TV', 'SB_TV', 'LLT_ROF', 'NDF', 'TGV', 'NLTV', 'Diff4th']"
    description:
        summary: The regularisation (denoising) method to stabilise the iterative recovery
        verbose: Iterative methods can help to solve ill-posed inverse problems by choosing a suitable noise model for the measurement
        options:
            ROF_TV: Rudin-Osher-Fatemi Total Variation model
            FGP_TV: Fast Gradient Projection Total Variation model
            PD_TV: Primal-Dual Total Variation
            SB_TV: Split Bregman Total Variation model
            LLT_ROF: Lysaker, Lundervold and Tai model combined with Rudin-Osher-Fatemi
            NDF: Nonlinear/Linear Diffusion model (Perona-Malik, Huber or Tukey)
            TGV: Total Generalised Variation
            NLTV: Non Local Total Variation
            Diff4th: Fourth-order nonlinear diffusion model
    default: FGP_TV

regularisation_parameter:
    visibility: basic
    dtype: float
    description:
        summary: Regularisation parameter. The higher the value, the stronger the smoothing effect
        range: Recommended between 1e-06 and 1e-04
    default: "5e-06"

regularisation_iterations:
    visibility: intermediate
    dtype: int
    description:
        summary: Total number of regularisation iterations. The smaller the number of iterations, the smaller the effect of the filtering is. A larger number will affect the speed of the algorithm.
        range: Recommended value dependent upon method.
    default:
        regularisation_method:
            ROF_TV: "300"
            FGP_TV: "100"
            PD_TV: "100"
            SB_TV: "100"
            LLT_ROF: "300"
            NDF: "300"
            Diff4th: "300"
            TGV: "150"
            NLTV: "30"
    dependency: regularisation_method

regularisation_PD_lip:
    visibility: advanced
    dtype: int
    description: Primal-dual parameter for convergence.
    default: "8"
    dependency:
        regularisation_method: PD_TV

regularisation_methodTV:
    visibility: advanced
    dtype: int
    description: 0/1 - TV specific isotropic/anisotropic choice.
    default: "0"
    dependency:
        regularisation_method:
            ROF_TV
            FGP_TV
            SB_TV
            NLTV

regularisation_timestep:
    visibility: advanced
    dtype: float
    dependency:
        regularisation_method:
            ROF_TV
            LLT_ROF
            NDF
            Diff4th
    description:
        summary: Time marching parameter
        range: Recommended between 0.0001 and 0.003
    default: "0.003"

regularisation_edge_thresh:
    visibility: advanced
    dtype: float
    dependency:
        regularisation_method:
            NDF
            Diff4th
    description:
        summary: Edge (noise) related parameter
    default: "0.01"

regularisation_parameter2:
    visibility: advanced
    dtype: float
    dependency:
        regularisation_method: LLT_ROF
    description:
        summary: Regularisation (smoothing) value
        verbose: The higher the value stronger the smoothing effect
    default: "0.005"

regularisation_NDF_penalty:
    visibility: advanced
    dtype: str
    options: "['Huber', 'Perona', 'Tukey']"
    description:
        summary: Penalty dtype
        verbose: Nonlinear/Linear Diffusion model (NDF) specific penalty type.
        options:
            Huber: Huber
            Perona: Perona-Malik model
            Tukey: Tukey
    dependency:
        regularisation_method: NDF
    default: Huber

Key

visibility: The visibility level of the parameter
dtype: The type of data
description: A short explanation of the parameter
default: The default value
options: A list of permitted values
dependency: The name of the parameter and value which this parameter depends upon
range: A guide for the range of the parameter