Algorithms#

AB_GMRES#

function [x,resL2,qualMeasOut]= AB_GMRES(proj,geo,angles,niter,varargin)
AB_GMRES solves the CBCT problem using AB_GMRES.
This is a stable Krylov method for when the backprojector is not adjoint

 AB_GMRES(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 AB_GMRES(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
'backprojector' Describes which backprojector to use

               *'FDK': This will use the FDK algorithm as a backprojector
                       NOTE: not the backprojector TIGRE calls "FDK", but
                       the actual algorithm
               *'matched': (DEFAULT) uses the pseudo-matched backprojectors

Source: MATLAB/Algorithms/AB_GMRES.m

ASD_POCS#

function [ f,qualMeasOut ] = ASD_POCS(proj,geo,angles,maxiter,varargin)
ASD_POCS Solves the ASD_POCS total variation constrained image in 3D
tomography.

  ASD_POCS(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  ASD_POCS(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda = lambdared * lambda. Default is 0.99

  'init':        Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)

                  •  'FDK'      : Initializes image to FDK reconstruction
  'TViter':      Defines the amount of TV iterations performed per SART
                 iteration. Default is 20

  'alpha':       Defines the TV hyperparameter. Default is 0.002

  'alpha_red':   Defines the reduction rate of the TV hyperparameter

  'Ratio':       The maximum allowed image/TV update ratio. If the TV
                 update changes the image more than this, the parameter
                 will be reduced. Default is 0.95
  'maxL2err'     Maximum L2 error to accept an image as valid. This
                 parameter is crucial for the algorithm, determines at
                 what point an image should not be updated further.
                 Default is 20% of the FDK L2 norm.
  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.

'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' : uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/ASD_POCS.m

AwASD_POCS#

function [ f,qualMeasOut]= AwASD_POCS(proj,geo,angles,maxiter,varargin)
AwASD_POCS Solves the 3D tomography problem using the adaptive-weighted
ASD_POCS algorithm which extends from the method ASD_POCS available in the
TIGRE toolbox by adding weight equation to better preserve the edge of the
reconstructed image

  AwASD_POCS(PROJ,GEO,ALPHA,NITER) added adaptive-weighted solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  AwASD_POCS(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)

Source: MATLAB/Algorithms/AwASD_POCS.m

AwPCSD#

function [ f,resL2,qualMeasOut] = AwPCSD(proj,geo,angles,maxiter,varargin)
AwPCSD solves the reconstruction problem using adaptive-weighted
projection-controlled steepest descent method

  AwPCSD(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem using
  the projection data PROJ taken over ALPHA angles, corresponding to the
  geometry described in GEO, using NITER iterations.

  AwPCSD(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros(default)

Source: MATLAB/Algorithms/AwPCSD.m

BA_GMRES#

function [x,resL2,qualMeasOut]= BA_GMRES(proj,geo,angles,niter,varargin)
BA_GMRES solves the CBCT problem using BA_GMRES.
This is a stable Krylov method for when the backprojector is not adjoint

 BA_GMRES(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 BA_GMRES(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
'backprojector' Descrives which backprojector to use

               *'FDK': This will use the FDK algorithm as a backprojector
                       NOTE: not the backprojector TIGRE calls "FDK", but
                       the actual algorithm
               *'matched': (DEFAULT) uses the pseudo-matched backprojectors

Source: MATLAB/Algorithms/BA_GMRES.m

B_ASD_POCS_beta#

function [ fres, qualMeasOut ] = B_ASD_POCS_beta(proj,geo,angles,maxiter,varargin)
B_ASD_POCS_beta Solves the ASD_POCS total variation constrained image in 3D
tomography using bregman iteration for the data.

  B_ASD_POCS_beta(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  B_ASD_POCS_beta(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every  iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros(default)

Source: MATLAB/Algorithms/B_ASD_POCS_beta.m

CGLS#

function [x,resL2,qualMeasOut]= CGLS(proj,geo,angles,niter,varargin)
CGLS solves the CBCT problem using the conjugate gradient least
squares

 CGLS(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 CGLS(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.

 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
 'restart'  true or false. By default the algorithm will restart when
            loss of orthogonality is found.

Source: MATLAB/Algorithms/CGLS.m

FBP#

function [res,errorL2]=FBP(proj,geo,angles,varargin)
TODO docs FBP

Source: MATLAB/Algorithms/FBP.m

FDK#

function [res]=FDK(proj,geo,angles,varargin)
FDK solves Cone Beam CT image reconstruction using Feldkam Davis Kress
algorithm (filtered backprojection)

  FDK(PROJ,GEO,ANGLES) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO.

  FDK(PROJ,GEO,ANGLES,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

  'parker': adds parker weights for limited angle scans. Default TRUE

  'wang': adds detector offset weights. Default TRUE

  'filter': selection of filter. Default 'ram-lak' (ramp)
             options are:
                 'ram-lak' (ramp)
                 'shepp-logan'
                 'cosine'
                 'hamming'
                 'hann'

Source: MATLAB/Algorithms/FDK.m

FISTA#

function [res,resL2,qualMeasOut] = FISTA(proj,geo,angles,niter,varargin)
FISTA is a quadratically converging algorithm, modified from FISTA.

It is based on the lazy-start FISTA modification in the following work:
J. Liang and C. Schonlieb, "Faster FISTA", in 26th European Signal
Processing Conference, 2018, pp.732-736. Lazy-start FISTA_mod

Source: MATLAB/Algorithms/FISTA.m

IRN_TV_CGLS#

function [x_outer,resL2,qualMeasOut]= IRN_TV_CGLS(proj,geo,angles,niter,varargin)
IRN_TV_CGLS solves the IRN_TV_CGLS problem using the conjugate gradient least
squares with Total Variation regularization, using an inner outer scheme

 IRN_TV_CGLS(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 IRN_TV_CGLS(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

 'lambda'  Value of regularization parameter lambda, default 10.
 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
 'restart'  true or false. By default the algorithm will restart when
            loss of orthogonality is found.

Source: MATLAB/Algorithms/IRN_TV_CGLS.m

LSMR#

function [x,resL2,qualMeasOut]= LSMR(proj,geo,angles,niter,varargin)
LSMR solves the CBCT problem using LSMR.

 LSMR(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 LSMR(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

 'lambda'  Value of parameter lambda, default 0.
 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
 'restart'  true or false. By default the algorithm will restart when
            loss of orthogonality is found.

Source: MATLAB/Algorithms/LSMR.m

LSQR#

function [x,resL2,qualMeasOut]= LSQR(proj,geo,angles,niter,varargin)
LSQR solves the CBCT problem using LSQR.
This is mathematically equivalent to CGLS.

 LSQR(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 LSQR(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
 'restart'  true or false. By default the algorithm will restart when
            loss of orthogonality is found.

Source: MATLAB/Algorithms/LSQR.m

MLEM#

function [res,qualMeasOut]=MLEM(proj,geo,angles,niter,varargin)
MLEM solves the tomographic problem by using Maximum Likelihood Expectation
Maximisation algorithm.

  MLEM(PROJ,GEO,ALPHA,NITER,opt) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

'verbose': Get feedback or not. Default: 1

'init':    Describes different initialization techniques.
            •  'none'     : Initializes the image to ones (default)
            •  'FDK'      : Initializes image to FDK reconstruction

'QualMeas':    Asks the algorithm for a set of quality measurement
               parameters. Input should contain a cell array of desired
               quality measurement names. Example: {'CC','RMSE','MSSIM'}
               These will be computed in each iteration.

'groundTruth':  An image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/MLEM.m

OSEM#

function [res,resL2,qualMeasOut]=OSEM(proj,geo,angles,niter,varargin)
OS_SART solves Cone Beam CT image reconstruction using Oriented Subsets
             Simultaneous Algebraic Reconstruction Technique algorithm

  OS_SART(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  OS_SART(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

  'BlockSize':   Sets the projection block size used simultaneously. If
                  if BlockSize = length(alpha) then OSEM becomes MLEM. Default is 20.

  'lambda':      Sets the value of the hyperparameter. Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda=lambdared*lambda. Default is 1

  'Init':        Describes different initialization techniques.
                 'none'     : Initializes the image to zeros (default)
                 'FDK'      : Initializes image to FDK reconstruction
                 'multigrid': Initializes image by solving the problem in
                              small scale and increasing it when relative
                              convergence is reached.
                 'image'    : Initialization using a user specified
                              image. Not recommended unless you really
                              know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' : uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
OUTPUTS:

   [img]                       will output the reconstructed image
   [img,errorL2]               will output the L2 norm of the residual
                               (the function being minimized)
   [img,errorL2,qualMeas]      will output the quality measurements asked
                               by the input 'QualMeas'

Source: MATLAB/Algorithms/OSEM.m

OS_ASD_POCS#

function [ fres, qualMeasOut ] = OS_ASD_POCS (proj,geo,angles,maxiter,varargin)
ASD_POCS Solves theOS_ASD_POCS total variation constrained image in 3D
tomography.

  OS_ASD_POCS(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  OS_ASD_POCS(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

  'Init':        Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)
                  •  'FDK'      : initializes image to FDK reconstruction
                  •  'image'    : Initialization using a user specified
                                  image. Not recommended unless you really
                                  know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.
  'TViter':      Defines the amount of TV iterations performed per SART
                 iteration. Default is 20

  'alpha':       Defines the TV hyperparameter. Default is 0.002

  'alpha_red':   Defines the reduction rate of the TV hyperparameter

  'Ratio':       The maximum allowed image/TV update ratio. If the TV
                 update changes the image more than this, the parameter
                 will be reduced. Default is 0.95
  'maxL2err'     Maximum L2 error to accept an image as valid. This
                 parameter is crucial for the algorithm, determines at
                 what point an image should not be updated further.
                 Default is 20% of the FDK L2 norm.
  'BlockSize':   Sets the projection block size used simultaneously. If
                 BlockSize = 1 OS-SART becomes SART and if  BlockSize = length(alpha)
                 then OS-SART becomes SIRT. Default is 20.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' :uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/OS_ASD_POCS.m

OS_AwASD_POCS#

function [f,qualMeasOut]= OS_AwASD_POCS(proj,geo,angles,maxiter,varargin)
OS_AwASD_POCS Solves the 3D tomography problem using the adaptive-weighted
OS_ASD_POCS algorithm which extends from the method OS_ASD_POCS available in the
TIGRE toolbox by adding weight equation to better preserve the edge of the
reconstructed image

  OS_AwASD_POCS(PROJ,GEO,ALPHA,NITER) added adaptive-weighted solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  OS_AwASD_POCS(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

  'Init':        Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)
                  •  'FDK'      : initializes image to FDK reconstruction

  'TViter':      Defines the amount of TV iterations performed per SART
                 iteration. Default is 20

  'alpha':       Defines the TV hyperparameter. Default is 0.002

  'alpha_red':   Defines the reduction rate of the TV hyperparameter

  'Ratio':       The maximum allowed image/TV update ratio. If the TV
                 update changes the image more than this, the parameter
                 will be reduced. Default is 0.95
  'maxL2err'     Maximum L2 error to accept an image as valid. This
                 parameter is crucial for the algorithm, determines at
                 what point an image should not be updated further.
                 Default is 20% of the FDK L2 norm.
  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
  'delta'        Defines Parameter to control the amount of smoothing
                 for pixels at the edges. A large 'delta' is not able to
                 differentiate image gradients at different pixels. A
                 small 'delta' give low weights to almost every pixels,
                 making the algorithm inefficient in removing noise or
                 straking artifacts. Default is -0.00055
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
  'BlockSize':   Sets the projection block size used simultaneously. If
                 BlockSize = 1 OS-SART becomes SART and if  BlockSize = length(alpha)
                 then OS-SART becomes SIRT. Default is 20.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' :uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/OS_AwASD_POCS.m

OS_AwPCSD#

function [ f,resL2,qualMeasOut] = OS_AwPCSD(proj,geo,angles,maxiter,varargin)
OS_AwPCSD solves the reconstruction problem using adaptive-weighted
projection-controlled steepest descent method

  OS_AwPCSD(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem using
  the projection data PROJ taken over ALPHA angles, corresponding to the
  geometry described in GEO, using NITER iterations.

  OS_AwPCSD(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros(default)

Source: MATLAB/Algorithms/OS_AwPCSD.m

OS_PCSD#

function [ f,resL2,qualMeasOut] = OS_PCSD(proj,geo,angles,maxiter,varargin)
OS_PCSD solves the reconstruction problem using adaptive-weighted
projection-controlled steepest descent method

  OS_PCSD(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem using
  the projection data PROJ taken over ALPHA angles, corresponding to the
  geometry described in GEO, using NITER iterations.

  OS_PCSD(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros(default)

Source: MATLAB/Algorithms/OS_PCSD.m

OS_SART#

function [res,resL2,qualMeasOut]=OS_SART(proj,geo,angles,niter,varargin)
OS_SART solves Cone Beam CT image reconstruction using Oriented Subsets
             Simultaneous Algebraic Reconstruction Technique algorithm

  OS_SART(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  OS_SART(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

  'BlockSize':   Sets the projection block size used simultaneously. If
                 BlockSize = 1 OS-SART becomes SART and if BlockSize = length(alpha)
                 then OS-SART becomes SIRT. Default is 20.

  'lambda':      Sets the value of the hyperparameter. Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda=lambdared*lambda. Default is 1

  'Init':        Describes different initialization techniques.
                 'none'     : Initializes the image to zeros (default)
                 'FDK'      : Initializes image to FDK reconstruction
                 'multigrid': Initializes image by solving the problem in
                              small scale and increasing it when relative
                              convergence is reached.
                 'image'    : Initialization using a user specified
                              image. Not recommended unless you really
                              know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' : uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.
OUTPUTS:

   [img]                       will output the reconstructed image
   [img,errorL2]               will output the L2 norm of the residual
                               (the function being minimized)
   [img,errorL2,qualMeas]      will output the quality measurements asked
                               by the input 'QualMeas'

Source: MATLAB/Algorithms/OS_SART.m

PCSD#

function [ f,qualMeasOut] = PCSD(proj,geo,angles,maxiter,varargin)
PCSD solves the reconstruction problem using projection-controlled steepest descent method

  PCSD(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem using
  the projection data PROJ taken over ALPHA angles, corresponding to the
  geometry described in GEO, using NITER iterations.

  PCSD(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

      'init':    Describes different initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)
                  •  'FDK'      : initializes image to FDK reconstruction

  'TViter':      Defines the amount of TV iterations performed per SART
                 iteration. Default is 20

  'maxL2err'     Maximum L2 error to accept an image as valid. This
                 parameter is crucial for the algorithm, determines at
                 what point an image should not be updated further.
                 Default is 20% of the FDK L2 norm.
  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/PCSD.m

PICCS#

function [ f,qualMeasOut ] = PICCS(proj,geo,angles,maxiter,prior,varargin)
PICCS Solves the PICCS total variation constrained image in 3D
tomography.

  PICCS(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry descrived in GEO, using NITER iterations.

  PICCS(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter for the SART iterations.
                 Default is 1

  'lambda_red':  Reduction of lambda.Every iteration
                 lambda=lambdared*lambda. Default is 0.99

  'init':        Describes diferent initialization techniques.
                  •  'none'     : Initializes the image to zeros (default)

Source: MATLAB/Algorithms/PICCS.m

SART#

function [res,resL2,qualMeasOut]=SART(proj,geo,angles,niter,varargin)
SART solves Cone Beam CT image reconstruction using Oriented Subsets
             Simultaneous Algebraic Reconstruction Technique algorithm

  SART(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  SART(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter. Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda=lambdared*lambda. Default is 1

  'Init':        Describes different initialization techniques.
                 'none'     : Initializes the image to zeros (default)
                 'FDK'      : Initializes image to FDK reconstruction
                 'multigrid': Initializes image by solving the problem in
                              small scale and increasing it when relative
                              convergence is reached.
                 'image'    : Initialization using a user specified
                              image. Not recommended unless you really
                              know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' : uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/SART.m

SART_TV#

function [res,resL2,qualMeasOut]=SART_TV(proj,geo,angles,niter,varargin)
SART_TV solves Cone Beam CT image reconstruction using Oriented Subsets
             Simultaneous Algebraic Reconstruction Technique algorithm

  SART_TV(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  SART_TV(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter. Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda=lambdared*lambda. Default is 0.99

  'Init':        Describes different initialization techniques.
                 'none'     : Initializes the image to zeros (default)
                 'FDK'      : Initializes image to FDK reconstruction
                 'multigrid': Initializes image by solving the problem in
                              small scale and increasing it when relative
                              convergence is reached.
                 'image'    : Initialization using a user specified
                              image. Not recommended unless you really
                              know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.

  'TViter'       number of iterations in the TV step. Default 50

  'TVlambda'     hyperparameter in TV iteration. It gives the ratio of
                 importance of the image vs the minimum total variation.
                 default is 15. Lower means more TV denoising.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.

  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
'OrderStrategy'  Chooses the subset ordering strategy. Options are
                 'ordered' : uses them in the input order, but divided
                 'random'  : orders them randomly
                 'angularDistance': chooses the next subset with the
                                    biggest angular distance with the ones used.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/SART_TV.m

SIRT#

function [res,resL2,qualMeasOut]=SIRT(proj,geo,angles,niter,varargin)
SIRT solves Cone Beam CT image reconstruction using Oriented Subsets
             Simultaneous Algebraic Reconstruction Technique algorithm

  SIRT(PROJ,GEO,ALPHA,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ALPHA angles, corresponding
  to the geometry described in GEO, using NITER iterations.

  SIRT(PROJ,GEO,ALPHA,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


  'lambda':      Sets the value of the hyperparameter. Default is 1

  'lambda_red':  Reduction of lambda. Every iteration
                 lambda=lambdared*lambda. Default is 1

  'Init':        Describes different initialization techniques.
                 'none'     : Initializes the image to zeros (default)
                 'FDK'      : Initializes image to FDK reconstruction
                 'multigrid': Initializes image by solving the problem in
                              small scale and increasing it when relative
                              convergence is reached.
                 'image'    : Initialization using a user specified
                              image. Not recommended unless you really
                              know what you are doing.
  'InitImg'      an image for the 'image' initialization. Avoid.

  'Verbose'      1 or 0. Default is 1. Gives information about the
                 progress of the algorithm.
  'QualMeas'     Asks the algorithm for a set of quality measurement
                 parameters. Input should contain a cell array of desired
                 quality measurement names. Example: {'CC','RMSE','MSSIM'}
                 These will be computed in each iteration.
'redundancy_weighting': true or false. Default is true. Applies data
                        redundancy weighting to projections in the update step
                        (relevant for offset detector geometry)
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/SIRT.m

hybrid_LSQR#

function [x,resL2, qualMeasOut,lambda_vec]= hybrid_LSQR(proj,geo,angles,niter,varargin)
hybrid_LSQR solves the CBCT problem using LSQR.

 hybrid_LSQR(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 hybrid_LSQR(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:

 'lambda'  Value of parameter lambda, default autocomputed.
 'Noiselevel' the expected nosie level, in %, replaces lambda.
 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.
 'groundTruth'  an image as ground truth, to be used if quality measures
                are requested, to plot their change w.r.t. this known
                data.

Source: MATLAB/Algorithms/hybrid_LSQR.m

hybrid_fLSQR_TV#

function [x,errorL2, qualMeasOut,lambda_vec]= hybrid_fLSQR_TV(proj,geo,angles,niter,varargin)
LSQR solves the CBCT problem using LSQR.
This is mathematically equivalent to CGLS.

 LSQR(PROJ,GEO,ANGLES,NITER) solves the reconstruction problem
  using the projection data PROJ taken over ANGLES angles, corresponding
  to the geometry described in GEO, using NITER iterations.

 LSQR(PROJ,GEO,ANGLES,NITER,OPT,VAL,...) uses options and values for solving. The
  possible options in OPT are:


 'Init'    Describes different initialization techniques.
            * 'none'     : Initializes the image to zeros (default)
            * 'FDK'      : initializes image to FDK reconstruction
            * 'multigrid': Initializes image by solving the problem in
                           small scale and increasing it when relative
                           convergence is reached.
            * 'image'    : Initialization using a user specified
                           image. Not recommended unless you really
                           know what you are doing.
 'InitImg'    an image for the 'image' initialization. Avoid.

Source: MATLAB/Algorithms/hybrid_fLSQR_TV.m