-
x.x.x
- New features:
- Added
Huber Lossfunction (#2281) AcquisitionGeometry.get_anglesreturns the angles in the requested unit, optionally including the initial angle (#2368)AcquisitionGeometry.set_initial_angleupdates the initial angle without re-setting the angle data (#2368)
- Added
- Enhancements:
GenericFilteredBackProjection'splot_filterreturns amatplotlib.figure.Figureinstead of amatplotlib.pyplot(#2360)TransmissionAbsorptionConvertersupportsDataContainers(#2372)
- Documentation:
- Corrected the
LADMMdocstring and user guide: the objective isf(x) + g(Kx)(notf(Kx) + g(x)), with fixed iteration formulas and a note on its equivalence toPDHG(#2367)
- Corrected the
- Bug fixes:
LSQRnow documents and warns that a non-zero scalaralpharegularises towardsinitial, minimising||Ax-b||^2 + alpha^2||x-initial||^2rather than||Ax-b||^2 + alpha^2||x||^2(#2357)- Fix
Parallel2Dsystem_descriptionraising exception for near-zero vectors (#2316) AcquisitionGeometry'sget_slicenow correctly checks Geometry type (#2378)
- New features:
-
26.0.0
- New features:
LSQRalgorithm added to the CIL algorithm class (#1975)- Add
VolumeShrinkertool to reduce the size of the reconstruction volume from anAcquisitionData(#2221) LaminographyGeometryCorrectortool added to processors (#2259)FluxNormalisercan be used onCone3D_Flexdata (#2347)
- Enhancements:
- Add prefix argument to
TIFFStackReaderto load a subset of TIFF files in a folder (#2239) - Update ASTRA interface to
direct_FP3D/BP3Dremoving copies for GPUProjectionOperatorcalls (#2134)
- Add prefix argument to
- Bug fixes:
CentreOfRotationCorrector.image_sharpnessdata is now correctly smoothed to reduce aliasing artefacts and improve robustness. (#2202)PaganinProcessornow correctly applies scaling with magnification for cone-beam geometry (#2225)cilaccpath lookup no longer broken for editable installations (#2257)- update
version.pyto useimportlib& fix tagless installation #2255 (#2269) - Fixed behaviour of
ZeissDataReaderwhen negative values are passed in the ROI (#2244) - Fix
show2Dtruncating plots when count is not a multiple ofnum_cols(#2315)
- Dependencies:
olefileanddxchangeare optional dependencies, instead of required (#2209, #2321)- fix missing
packagingrequired dependency (#2321) - dxchange minimum version set to 0.2.1 to fix #2256 (#2268)
- improve
tqdmnotebook support (#2241) cvxpyversion set to !=1.8.2 to fix #2303 (#2306)- Handle regularisation toolkit CPU only package error message (#2302)
- Update
FindIPP.cmaketo find IPP libraries in conda environments (#2286) - Update to ASTRA-TOOLBOX version v2.4 from the
astra-toolboxchannel (#2330) - Update to TIGRE v3.1.3 (#2317)
- Added support for numpy 2 (#2218)
- Added utililty method
dtype_liketo convert an input value to the same type as a reference array - Added documentation to developer guide on scalar types with numpy>=2.
- Added utililty method
- Update to CCPi-Regularisation toolkit v26.0.0 (#2218)
- Update to Tomophantom v3.1.4 (#2218)
- Added support for python 3.13 (#2218)
- Documentation:
- Render the user showcase notebooks in the documentation (#2189)
- Update on build instructions in README and developer guide for all OS (#2286)
- Breaking Changes:
- CIL is no longer compatible with Tomophantom versions earlier than v3.1.4 (#2218)
- Removes the following code which had been deprecated since v24.3.0 or earlier:
- Removes
max_iterationandlog_fileinput parameters toAlgorithms. - Removes
max_iteration_stop_criterion,objective_to_string,verbose_outputandverbose_headermethods fromAlgorithms. - Removes
print_intervalandcallbackkwargs fromAlgorithm.run(note:callbackskwarg remains) - Removes
toleranceinput parameter toCGLS. - Removes
should_stopandflagmethods fromCGLS. - Removes
alpha,beta,rtolandatolkwargs fromGD. - Removes
should_stopandobjective_functionmethods fromGD. - Removes
normsandprobkwargs fromSPDHG.
- Removes
- Build changes:
- migrate from
conda buildtorattler-build(#2321)
- migrate from
- New features:
-
25.0.0
- New features:
- Added
FunctionOfAbsclass (#1976) - Added
Cone3D_Flexgeometry as a CIL acquisition geometry type. This allows users to reconstruct non-circular trajectories. (#2039) - Added
ASTRAbackend support forProjectionOperatorandFBPto useCone3D_Flexgeometry. (#2039) - Added
show_system_positionsfor visualisation ofCone3D_Flexgeometries. (#2039) - Added an Accelerated Proximal Gradient Descent Algorithm (APGD) with options for momentum (#2145)
- Added
- Enhancements:
- Added an accelerated version to the TransmissionAbsorption processor, controlled by the
acceleratedparameter (default isTrue) (#2036) - Made the call to next() in algorithm iteration loop explicit (#2069)
- Added option for a random seed in the power method in the linear operator (#1585)
- Improved efficiency of
Normaliserprocessor. Reduced memory use and increased speed (#2111) - Extra functionality for sampler:
get_previous_samples()andget_current_sample()(#2079) - PDHG 'check_convergence' updated for new literature (#2084)
- Updated
create_local_env_for_cil_development.shscript to work on Windows (#2144) - Improved consistency of
step_sizeproperty across GD, ISTA, FISTA and APGD algorithms (#2157) - In PDHG algorithm, we now have options to initialise the dual variable, as well as the primal variable (#2169)
- Allow FluxNormaliser.preview_configuration() even if flux contains zeros (#2177)
- Check if kwargs are used in AcquisitionData and ImageData initialisation (#2178)
- Added a flag to
show_geometrythat allows for disabling the call toplt.show()(#2195) - The random methods to populate a data container can now be accessed via
fill()andallocate(), and will have a lower memory footprint on creation (#2037)
- Added an accelerated version to the TransmissionAbsorption processor, controlled by the
- Bug fixes:
- Fix deprecation warning for rtol and atol in GD (#2056)
- Removed the deprecated usage of run method in test_SIRF.py (#2070)
- Ensured CIL forward and back projectors always return, even when
outis passed (#2059) - Made ProjectionOperator
deviceinput case-insensitive (#1990) - Ensured the same approximate gradient is used within each iteration for PD3O with a stochastic function
f(#2043) - Fix
recon.FBPsplit_processingmethods forASTRAbackend (#2114) - Copy geometry in the creation of a DataContainer (#2108)
- Fix order of operations for subtraction and division between a BlockDataContainer and DataContainer (#2133)
- Update default compression for TIFFWriter setup (#2197)
- ProjectionOperator device input is no longer case-sensitive (#2065)
- Fix
pkg_resourcesdeprecation warning onsetuptools>=81 dxchange<=0.2.0(#2207)
- Deprecated
- Positional arguments for
get_sliceonAcquisitionGeometryandAcquisitionDatahave been deprecated (#2039) - Methods to allocate a
DataContainerusing the old numpy random number generator have been renamed and deprecated,RANDOM->RANDOM_DEPRECATEDandRANDOM_INT->RANDOM_INT_DEPRECATED. See "Breaking Changes" for details of new behaviour (#2037) - Sampler's
get_sampleshas been deprecated, useview_samples(#2128)
- Positional arguments for
- Removed code
- Removed the following code which has been deprecated since v23.0.0 or earlier (#2150):
dimension_labelskwarg inAcquisitionGeometry.allocate()and inImageData's__init__axpby(alias ofsapyb)shapesetter inDataContainerandImageGeometry- use of integer compression in
NEXUSDataWriter
- Removed unused kwargs passed to ImageData and AcquisitionData, including
suppress_warning(#2206)
- Removed the following code which has been deprecated since v23.0.0 or earlier (#2150):
- Documentation:
- Updated documentation for the ChannelWiseOperator including new example (#2096)
- Updated documentation for LADMM (#2015)
- Updated Contributor's Guide to include an example of parametrized tests, using
unittest-parametrize(#1990) - Tidied up documentation of
L1Norm(#2186) - Added callbacks to documentation (#2067)
- Added new tutorials to the "How-To" page in the docs (#1972)
- Testing:
- Developers can now run the full CI matrix via the web UI (#2160)
- Added tests for ProjectionOperator inputs that use
unittest-parametrizemodule (#1990) - Added tests for Normaliser processor
- Update minimum cmake version to 3.5
- Added unit test for the
ZEISSDataReader(#2098) - Added
requirements-test-windows.ymlto create test environment (#2174) - Automate Windows conda build & release (#1918)
- Dependencies:
- Move from CMake to
pip install(#2145) - Use conda native compilers (#2199)
- matplotlib-base is an optional dependency, instead of required (#2093)
unittest-parametrizehas been added as a dependency for tests (#1990)- Defer import of olefile and dxchange (#2149)
- zenodo_get is an optional dependency, instead of required (#2146)
FindIPP.cmakefallback (#2148)- enable
find_packageusing*_ROOThints - simplified
cilaccbuild logic for optional dependencies
- enable
- Removed upper version constraints on optional plugins when resolving conda environment (#2204)
- Relaxed numpy runtime pinning of versions (#2156)
- Move from CMake to
- Breaking changes:
RANDOMandRANDOM_INToptions inDataContainer.fill()andgeometry.allocate()methods now use numpy default random number generator to populate the array. Values for a set seed will now change compared to previous versions. Previous methods can be accessed withRANDOM_DEPRECATEDandRANDOM_INT_DEPRECATED. (#2037)RANDOMandRANDOM_INToptions inDataContainer.fill()andgeometry.allocate()methods will no longer use a random seed set externally usingnumpy.random.seed(), the seed should be provided as an argument.RANDOM_DEPRECATEDandRANDOM_INT_DEPRECATEDwill still use an externally set seed (#2037)
- New features:
-
24.3.0
- New features:
- Added
FluxNormaliserprocessor (#1878)
- Added
- Bug fixes:
- Fix bug with 'median' and 'mean' methods in Masker averaging over the wrong axes (#1548)
SPDHGgammaparameter is now applied correctly so that the product of the dual and primal step sizes remains constant asgammavaries (#1644)- Allow MaskGenerator to be run on DataContainers (#2001)
- Make
PaganinProcessorwork withAcquistionDatawith one angle (#1920) - Fix bug passing
kwargsto PDHG (#2010) show1Dcorrectly applies slices to N-dimensional data (#2022)BlockOperatordirect and adjoint methods: can pass out as aDataContainerinstead of a (1,1)BlockDataContainerwhere geometry permits (#1926)- Add path to python executable to cmake commands to fix issue with cmake retrieving wrong python (#2044)
- Enhancements:
- Removed multiple exits from numba implementation of
KullbackLeiblerdivergence (#1901) - Updated the
SPDHGalgorithm to take a stochasticSampler(#1644) - Updated the
SPDHGalgorithm to include setters for step sizes (#1644) - SAPBY for the
BlockDataContainernow does not require anoutto be passed (#2008) - Fixed the rendering of the SAG/SAGA documentation (#2011)
- Set aliases: ISTA=PGD, FISTA=APGD (#2007)
- Removed multiple exits from numba implementation of
- Dependencies:
- Added scikit-image to CIL-Demos conda install command as needed for new Callbacks notebook (#1955)
- Replaced matplotlib dependency with matplotlib-base (#2031)
- Remove CIL-Data from build requirements, update version to >=22 in run requirements (#2046)
- Changes that break backwards compatibility:
show1Dargument renamedlabel->dataset_labels, default plot size has changed. (#2022)show1DDefault behaviour for displaying and labeling multiple plots has changed. Each slice requested will be displayed on a new subplot comparing all datasets at that position. (#2022)- The
runmethod in the cil algorithm class will no longer run if a number of iterations is not passed (#1940) - Paganin processor now requires the CIL data order (#1920)
- The gradient descent algorithm now takes
finstead ofobjective_functionto match with ISTA and FISTA (#2006)
- Deprecated code:
- Deprecated
normsandprobin theSPDHGalgorithm to be set in theBlockOperatorandSamplerrespectively (#1644) - Deprecated
rtolandatolfrom GD so that it does not stop iterating automatically - for this functionality users should use a callback (#1944)
- Deprecated
- Testing:
- Added a new test file
test_algorithm_convergencethat will hold our algorithm tests that run to convergence (#2019) - Update random array in test_stepsizes.TestStepSizeBB.test_bb_converge to fix tests on windows (#2052)
- Added a new test file
- New features:
-
24.2.0
- New Features:
- Added SVRG and LSVRG stochastic functions (#1625)
- Added SAG and SAGA stochastic functions (#1624)
- Allow
SumFunctionwith 1 item (#1857) - Added PD3O algorithm (#1834)
- Added Barzilai-Borwein step size rule to work with GD, ISTA, FISTA (#1859)
- Added callback
optimisation.utilities.callbacks.EarlyStoppingObjectiveValuewhich stops iterations if an algorithm objective changes less than a provided threshold (#1892) - Added callback
optimisation.utilities.callbacks.CGLSEarlyStoppingwhich replicates the automatic behaviour of CGLS in CIL versions <=24. (#1892) - Added
labelsmodule withImageDimension,AcquisitionDimension,AcquisitionType,AngleUnit,FillType(#1692)
- Enhancements:
- Use ravel instead of flat in KullbackLeibler numba backend (#1874)
- Upgrade Python wrapper (#1873, #1875)
- Updated the documentation for the algorithm base class (#1809)
- Add checks on out argument passed to processors to ensure correct dtype and size (#1805)
- Internal refactor: Replaced string-based label checks with enum-based checks for improved type safety and consistency (#1692)
- Internal refactor: Separate framework into multiple files (#1692)
- Allow the SIRT algorithm to take
initial=None(#1906) - Add checks on equality method of
AcquisitionDataandImageDatafor equality of data type and geometry (#1919) - Add check on equality method of
AcquisitionGeometryfor equality of dimension labels (#1919)
- Testing:
- New unit tests for operators and functions to check for in place errors and the behaviour of
out(#1805) - Updates in SPDHG vs PDHG unit test to reduce test time and adjustments to parameters (#1898)
- Drop Jenkins in favour of GHA for conda builds (#1914)
- New unit tests for
DataContainer,AcquisitionDataandImageDatato check equality method (__eq__) behaves as expected (#1919)
- New unit tests for operators and functions to check for in place errors and the behaviour of
- Bug fixes:
ImageDataremoves dimensions of size 1 from the input array. This fixes an issue where single slice reconstructions from 3D data would fail due to shape mismatches (#1885)- Make Binner accept accelerated=False (#1887)
- Added checks on memory allocations within
FiniteDifferenceLibrary.cppand verified the status of the return inGradientOperator(#1929) - Build release version of
cilacc.dllfor Windows. Previously was defaulting to the debug build (#1928) - Armijo step size rule now by default initialises the search for a step size from the previously calculated step size (#1934)
- Changes that break backwards compatibility:
- CGLS will no longer automatically stop iterations once a default tolerance is reached. The option to pass
tolerancewill be deprecated to be replaced byoptimisation.utilities.callbacks(#1892)
- CGLS will no longer automatically stop iterations once a default tolerance is reached. The option to pass
- New Features:
-
24.1.0
- New Features:
- Added method to plot filter in
GenericFilteredBackProjection(#1667) - Added wavelet operator, wrapping PyWavelets operator as a CIL operator (#1615)
- Added PaganinProcessor processor, to perform phase retrieval from phase contrast images (#1737)
- Added L1Sparsity function, allowing calculations of
|Ax-b|_1and its proximal, in the case of orthogonal operators,A(#1618) - Options in algorithms GD, ISTA and FISTA to pass a
cil.optimisation.utilities.StepSizeRuleor acil.optimisation.utilities.Preconditioner(#1768) - an implementation of the Armijo Rule as a child class of
cil.optimisation.utilities.StepSizeRule(#1768) - Sensitivity preconditioners added as child classes of
cil.optimisation.utilities.Preconditioner(#1768)
- Added method to plot filter in
- Enhancements:
- Added
geometryproperty toBlockDataContainer. Adds__eq__toBlockGeometry(#1799) - Raises error in
BlockDataContainer.pnormif the shape of the containers is not the same (#1799) - Operators and functions now also return when out is specified (#1742)
- The CIL function class now has a
__neg__function, so you can write-YourFunction(x)rather than-1*YourFunction(x)(#1808) - Added documentation for the Partitioner to
framework.rst(#1828) - Added CIL vs SIRF tests comparing preconditioned ISTA in CIL and MLEM in SIRF (#1823)
- Update to CCPi-Regularisation toolkit v24.0.1 (#1868)
- Added
- Bug fixes:
- gradient descent
update_objectivecalled twice on the initial point.(#1789) - ProjectionMap operator bug fix in adjoint and added documentation (#1743)
- BlockOperator that would return a BlockDataContainer of shape (1,1) now returns the appropriate DataContainer. BlockDataContainer direct and adjoint methods accept DataContainer as parameter (#1802).
- BlurringOperator: remove check for geometry class (old SIRF integration bug) (#1807)
- The
ZeroFunctionandConstantFunctionnow have a Lipschitz constant of 1. (#1768) - Update dataexample remote data download to work with windows and use zenodo_get for data download (#1774)
- gradient descent
- Changes that break backwards compatibility:
- Merged the files
BlockGeometry.pyandBlockDataContainer.pyinframeworkto one fileblock.py. Please usefrom cil.framework import BlockGeometry, BlockDataContaineras before (#1799) - Bug fix in
FGP_TVfunction to set the default behaviour not to enforce non-negativity (#1826).
- Merged the files
- New Features:
-
24.0.0
- Update to new CCPi-Regularisation toolkit v24.0.0. This is a backward incompatible release of the toolkit.
- CIL plugin support for TIGRE version v2.6
- CIL plugin support for ASTRA-TOOLBOX version v2.1
- Dropped support for python 3.8 and 3.9
- Added support for python 3.11 and 3.12
- Dropped support for numpy 1.21 and 1.22
- Added support for numpy 1.25 and 1.26
- Set CMake Policy CMP0148 to OLD to avoid warnings in CMake 3.27
- AcquisitionGeometry prints the first and last 10 angles, or all if there are 30 or less, rather than the first 20
- Added a weight argument to the L1Norm function
- Allow reduction methods on the DataContainer class to accept axis argument as string which matches values in dimension_labels
- Added the functions
set_normsandget_normsto theBlockOperatorclass - Internal variable name change in BlockOperator to aid understanding
- Fixed formatting errors in the L2NormSquared and LeastSquares documentation
- Bug fix for BlockDataContainer as iterator
- Dropped support for IPP versions older than 2021.10 due to header changes
- Fix build include directories
- Proximal of MixedL21Norm with numpy backend now accepts numpy ndarray, DataContainer and float as tau parameter
- ZeroOperator no longer relies on the default of allocate
- Bug fix in SIRF TotalVariation unit tests with warm_start
- Allow show2D to be used with 3D
DataContainerinstances - Added the a
Samplerclass as a CIL optimisation utility - Update documentation for symmetrised gradient
- Added documentation for CompositionOperator and SumOperator
- Updated FISTA and ISTA algorithms to allow input functions to be None
- Bug fix in the adjoint of the Diagonal Operator for complex values
- Update conda build action to v2 for 2.5x quicker builds
- Add docker image, test & push to
ghcr.io/tomographicimaging/cil - Quality metrics have added mask option
- Bug fix for norm of CompositionOperator
- Functions updated to use sapyb
- Fix KullbackLeibler numba gradient ignoring mask
- show1D slice_list parameter can now be of type tuple
- Deprecated
Algorithm'smax_iteration,log_file,**kwargs,max_iteration_stop_criterion,objective_to_string,verbose_output,verbose_header,run(print_interval) - Added
optimisation.utilities.callbacks- Added
Callback(abstract base class),ProgressCallback,TextProgressCallback,LogfileCallback - Deprecated
Algorithm.run(callback: Callable) - Added
Algorithm.run(callbacks: list[Callback])
- Added
- Allow
CentreOfRotationCorrector.xcorrelationto accept a list of two values inprojection_indexto use for the correlation - Bug fix in
CentreOfRotationCorrector.xcorrelationfinding correlation angles for limited angle data - New unit tests have been implemented for operators and functions to check for in place errors and the behaviour of
out. - Bug fix for missing factor of 1/2 in SIRT update objective and catch in place errors in the SIRT constraint
- Bug fix to allow safe in place calculation for the soft shrinkage algorithm
- Allow Masker to take integer arrays in addition to boolean
- Add remote data class to example data to enable download of relevant datasets from remote repositories
- Improved import error/warning messages
- New adjoint operator
- Bug fix for complex matrix adjoint
- Removed the following code which was deprecated since v22.0.0:
infoparameter incil.optimisation.functions.TotalVariationsinogram_geometryandvolume_geometryparameters incil.plugins.astra.processors.FBPand incil.plugins.tigre.processors.FBPaquisition_geometry(misspelled) parameter incil.plugins.tigre.processors.ProjectionOperatorFBPkwarg (and thus all kwargs) incil.processors.CentreOfRotationCorrectorandcil.processors.CofR_image_sharpnessTXRMDataReader
- Added the ApproximateGradientSumFunction and SGFunction to allow for stochastic gradient algorithms to be created using functions with an approximate gradient and deterministic algorithms
-
23.1.0
- Fix bug in IndicatorBox proximal_conjugate
- Allow CCPi Regulariser functions for non CIL object
- Add norm for CompositionOperator
- Refactor SIRT algorithm to make it more computationally and memory efficient
- Optimisation in L2NormSquared
- Added support for partitioner, when partitions have size 1
- Fix for show_geometry bug for 2D data
- FBP split processing bug fix - now respects panel origin set in geometry
- Binner/Padder/Slicer bug fix - now respects panel origin set in geometry
- Added warmstart capability to proximal evaluation of the CIL TotalVariation function.
- Bug fix in the LinearOperator norm with an additional flag for the algorithm linearOperator.PowerMethod
- Tidied up documentation in the framework folder
-
23.0.1
- Fix bug with NikonReader requiring ROI to be set in constructor.
-
23.0.0
- Partitioner is now able to create batches even if angle is not the outer dimension
- Renamed
max_iteration_stop_cryterionmethod in the Algorithm class tomax_iteration_stop_criterion - Removed (previously deprecated)
very_verboseparameter inAlgorithm's run method. - Removed (previously deprecated)
axpbymethod in DataContainer. - Deprecate use of integer compression in NEXUSDataWriter.
- Improved and tidied up documentation for all readers and writers, including hiding special members.
- Use arguments instead of kwargs in all readers and writers with multiple kwargs, making documentation easier.
- Update Apache2 License Headers.
-
22.2.0
- BlockGeometry is iterable
- Added
partitiontoAcquisitionDatato partition the data with 3 methods:sequential,staggeredandrandom_permutation - TIGRE and ASTRA
ProjectionOperatornow supportBlockGeometryasacquisition_geometryparameter, returning aBlockOperator - Added pre-set filters for
recon.FBPandrecon.FDK. Filters now include ram-lak, hamming, hann, cosine, shepp-logan. - Added RAWFileWriter to export data containers to raw files
- Extended IndicatorBox to behave as IndicatorBoxPixelwise by passing masks in lower and upper bounds
- Implemented IndicatorBox in numba and numpy
- Dropped support for Python 3.6 and NumPy 1.15
- Jenkins PR tests on Python 3.8 and NumPy 1.20
- added yml file to create test environment
- LeastSquares fixed docstring and unified gradient code when out is passed or not.
- Add compression to 8bit and 16bit to TIFFWriter
- Added convenience centre of rotation methods to
AcquisitionGeometryclass.get_centre_of_rotation()calculates the centre of rotation of the systemset_centre_of_rotation()sets the system centre of rotation with an offset and angleset_centre_of_rotation_by_slice()sets the system centre of rotation with offsets from two slices
- Binner processor reworked:
- Significant speed increase available via the C++ backend
- Returned geometry is correctly offset where binning/cropping moves the origin
- Slicer refactoring
- Returned geometry is correctly offset where slicing/cropping moves the origin
- Padder refactoring
- Returned geometry is correctly offset where padding moves the origin
- Github Actions:
- update test python and numpy versions to 3.9 and 1.22
- Update conda build action to v1.4.4
- Fixes actions to run on ubuntu-20.04
- Update version of upload_artifact github action to version 3.1.1
- Update version of download_artifact github action to version 3.0.1
- Update version of checkout github action to version 3.1.0
- Update build-sphinx action to version 0.1.3
io.utilities.HDF5_utilitiesAdded utility functions to browse hdf5 files and read datasets into numpy array- Implemented the analytical norm for GradientOperator
- Added
ImageData.apply_circular_maskmethod to mask out detector edge artefacts on reconstructed volumes - ROI selection, aspect ratio toggle and Play widget added to islicer
- Add show1D display utility
-
22.1.0
- use assert_allclose in test_DataContainer
- added multiple colormaps to show2D
- Fix segfault in GradientOperator due to parameter overflows on windows systems
- Fix angle display precision and matplotlib warning for sinograms with show2D
-
22.0.0
- Strongly convex functionality in TotalVariation and FGP_TV Functions
- Refactored KullbackLeibler function class. Fix bug on gradient method for SIRF objects
- Numba added as a CIL requirement
- Simplify initialisation of
CentreOfRotation.ImageSharpnesswith new parameterbackend - Added ISTA algorithm. Improve inheritance of proximal gradient algorithms
- Updated interface to
plugins.tigre/plugins.astraFBPandProjectionOperatorclasses - Update NikonDataReader to parse and set up geometry with:
ObjectTiltCentreOfRotationTopandCentreOfRotationBottom - Cleaned up unit test structure and output
- Removal of deprecated code:
- AcquisitionGeometry
__init__no longer returns a configured geometry, use factorycreatemethods instead subsetmethod removed, useget_sliceorreordermethods- NikonDataReader
normalizekwarg removed, usenormalise - Algorithms initialisation
x_initkwarg removed, useinitial - Removed deprecated numpy calls
- AcquisitionGeometry
- DataProcessors use weak-reference to input data
- Merged CIL-ASTRA code in to CIL repository simplifying test, build and install procedures
- Modules not moved should be considered deprecated
- CIL remains licensed as APACHE-2.0
- Minor bug fixes to the CPU 2D Parallel-beam FBP
- Add ndim property for DataContainer class
- Fixes show_geometry compatibility issue with matplotlib 3.5
- Added ZEISSDataReader with cone/parallel beam, slicing, TXM Functionality
- Raise exception if filename or data haven't been set in NexusDataWriter
- Fixes error when update_objective_interval is set to 0 in an algorithm run
- Deprecated:
- TXRMDataReader is deprecated in favour of ZEISSDataReader
- GitHub Actions:
- Update to version 0.1.1 of lauramurgatroyd/build-sphinx-action for building the documentation - ensures docs are always built from cil master
-
21.4.1
- Removed prints from unittests and cleanup of unittest code.
- CMake:
- install script re-allows selection of non default install directory (#1246)
- TIFF writer uses logging
- Added unittests for TIFF functionality
-
21.4.0
- PEP 440 compliant version
- CMake fix due to use of pip install.
- Recon.FBP allows 'astra' backend
- Fixed PowerMethod for square/non-square, complex/float matrices with stopping criterion.
- CofR image_sharpness improved for large datasets
- Geometry alignment fix for 2D datasets
- CGLS update for sapyb to enable complex data, bugfix in use of initial
- added sapyb and deprecated axpby. All algorithm updated to use sapyb.
- Allow use of square brackets in file paths to TIFF and Nikon datasets
-
21.3.1
- Added matplotlib version dependency to conda recipe
- Fixed TIGRE wrappers for geometry with a virtual detector
- Fixed TIGRE wrappers for cone-beam geometry with tilted rotation axis
-
21.3.0
- Accelerated PDHG which handles strong convexity of functions
- TotalVariation Function handles SIRF ImageData
- Simulated datasets and volume added to DataExamples
- TIGRE wrappers for parallel-beam geometry added
- NEXUSWriter and NEXUSReader offer (8bit and 16bit) compression of data
- show2D show_geom now return an object that can be saved with a
savemethod - GradientOperator can be now used with SIRF DataContainers, both PET and MR
- Add anisotropy in TotalVariation function
- CCPi Regularisation plugin is refactored, only FGP_TV, FGP_dTV, TGV and TNV are exposed. Docstrings and functionality unit tests are added. Tests of the functions are meant to be in the CCPi-Regularisation toolkit itself.
- Add dtype for ImageGeometry, AcquisitionGeometry, VectorGeometry, BlockGeometry
- Fix GradientOperator to handle pseudo 2D CIL geometries
- Created recon module with FBP and FDK using fast filtering library and TIGRE backprojectors
- Added Intel IPP based library for filtering step of FBP
- PDHG memory optimisation
- ScaledFunction memory Optimisation
- The github actions are merged into one action with multiple jobs
- The conda build job uploads an artifact of the build tar.bz file which is later used by the documentation build job - which installs the package into a miniconda environment.
- Documentation pages for recon, astra and cil-plugins are published.
-
21.2.0
- add version string from git describe
- add CCPi-Regularisation toolkit in unittests
- show_geometry implemented to display AcquisitionGeometry objects, can be imported from utilities.display
- CentreOfRotationCorrector.image_sharpness implemented which finds the rotation axis offset by maximising sharpness of a single slice reconstruction
- Renamed CentreOfRotationCorrector.xcorr to CentreOfRotationCorrector.xcorrelation
- Implemented Padder processor
-
21.1.0
- Added TomoPhantom plugin to create 2D/3D + channel ImageData phantoms based on the TomoPhantom model library
- Fixed bug in Zeiss reader geometry direction of rotation
-
21.0.0
- Show2D now takes 4D datasets and slice information as input
- TIGRE reconstruction package wrapped for cone-beam tomography
- Datacontainers have get_slice method which returns a dataset with a single slice of the data
- Datacontainers have reorder method which reorders the data in memory as requested, or for use with 'astra' or 'tigre'
- Subset method has been deprecated
- AcquisitionData and ImageData enforce requirement for a geometry on creation
- New processors AbsorptionTransmissionConverter and TransmissionAbsorptionConverter to convert between Absorption and Transmission
- Implemented Binner and Slicer processors
- Implemented MaskGenerator and Masker processors
-
20.11.2
- fixed windows build
- NikonDataReader converts Nikon geometry to CIL geometry from xtekct file including detector and centre-or-rotation offsets
- NexusdataReader supports files written with old versions of NexusDataWriter
-
20.11
- python module renamed to cil
- renamed Identity->IdentityOperator, Gradient->GradientOperator, SymmetrisedGradient->SymmetrisedGradientOperator
-
20.09.1
- FiniteDifferenceOperator takes into consideration voxel size
- Added CentreOfRotationCorrector
- Removed CenterOfRotationFinder
- moved TestData to utilities and renamed as dataexample
- verbosity of Algorithms is independent of the update_objective_interval
- added unittests
- renamed
- GradientDescent to GD
- SparseFiniteDiff to SparseFiniteDifferenceOperator
- LinearOperatorMatrix to MatrixOperator
- bugfix update_objective of SPDHG
-
20.09
- added SPDHG algorithm
- added TotalVariation function
- Redesign of the AcquisitionGeometry class allowing more general acquisition trajectories than currently possible.
- Added ZEISS reader
-
20.04
- Significant upgrades to the operator and function classes to allow more flexible definition of optimisation problems
- Added multithreaded C library for calculation of finite difference and some data processing
- Added Gradient operator using C library and numpy backends
-
19.10
- Improved usability with reader/writers and plotting utilities
- Substantially improved test coverage
-
19.07
- Introduction of BlockFramework
- major revision and restructuring of the whole code
- rewritten io package
-
19.02
- introduction of Algorithm class
- unit test expanded and moved to test directory
- unified build system on Jenkins based on CCPi-VirtualMachine repo
- switched to calendar versioning YY.0M.
-
0.10.0
-
0.9.4
- Initial release