Awesome Computed Tomography

A curated list of CT reconstruction resources. Every paper link is resolved against Crossref, arXiv or the publisher’s own record, and every URL is rechecked weekly.

Papers: 678
Toolkits: 60
Datasets: 82
Benchmarks: 22
Learning: 35
Updated: 2026-10-30

CT Reconstruction Toolkits

60 open-source toolkits and libraries for CT reconstruction: reconstruction engines, differentiable operators, simulators and viewers, with language, licence and GitHub stars.

Reconstruction engines (20)

Toolkit	Language	License	Stars	Notes
TIGRE	MATLAB / Python	BSD-3-Clause	814	GPU-accelerated (CUDA) toolbox for cone-beam and parallel-beam CT; FDK, SART, SIRT, CGLS and TV variants.
ASTRA Toolbox [site]	C++ / Python / MATLAB	GPL-3.0	653	Long-standing research toolbox with a large set of projectors and algorithms; the reference CPU/GPU baseline.
TomoPy	Python	BSD-3-Clause	396	Reference toolkit for synchrotron micro/nano tomography; gridrec, FBP and phase retrieval.
RTK [site]	C++	Apache-2.0	298	Reconstruction Toolkit built on ITK; widely used in clinical CBCT and interventional research.
LEAP	CUDA / Python	MIT	257	LLNL's LivermorE AI Projector for Computed Tomography (LEAP); production-grade model-based iterative reconstruction for 3D transmission CT.
PyTomography	Python	MIT	178	End-to-end medical image reconstruction with quantification, including CT-based attenuation correction.
Core Imaging Library (CIL) [site]	Python	Apache-2.0	156	CCPi framework for variational CT reconstruction with a large catalogue of optimisation algorithms.
OMEGA	MATLAB / Python	GPL-3.0	111	Open-source multi-dimensional tomographic reconstruction software covering PET/SPECT/CT.
tomosipo	Python	GPL-3.0	107	Painless specification of 2D/3D projection geometries; wraps ASTRA and integrates with PyTorch.
DBT-Reconstruction	MATLAB	GPL-3.0	62	Open-source reconstruction toolbox for digital breast tomosynthesis, covering DBT-specific geometry alongside FBP and iterative solvers.
ToMoBAR [site]	Python	Apache-2.0 / GPL-3.0	54	Python library for model-based iterative reconstruction (FISTA, ADMM, ordered subsets) with the CCPi regularisers, built on ASTRA; developed at Diamond Light Source.
tomocupy [site]	Python	BSD-style (see repository)	30	GPU tomographic reconstruction from Argonne's Advanced Photon Source, built on CuPy for fast reconstruction of large synchrotron datasets.
svmbir	C / Python	BSD-3-Clause	28	Fast Super-Voxel Model Based Iterative Reconstruction with proximal-map plug-ins.
MBIRJAX [site]	Python	BSD-3-Clause	23	Model-based iterative reconstruction in JAX from Bouman's group, running on CPU or GPU; the successor to svmbir.
tofu	Python	LGPL-3.0	20	Helper layer on top of the UFO framework for fast GPU tomographic reconstruction pipelines.
libcbct	C++	CC BY-NC-SA 4.0	10	Lightweight cone-beam CT reconstruction library with FDK and iterative solvers; non-commercial licence.
mumott [site]	Python	GPL-3.0	4	Tensor-tomography reconstruction for reciprocal-space data, developed at DTU. Specialised rather than general-purpose, but it is the most complete open implementation of the tensor-tomography model.
DIRECTT Reconstruction GUI	Python	GPL-2.0	1	A Python GUI built on the ASTRA toolbox for reconstructing X-ray CT data, with the DIRECTT algorithm included. DIRECTT targets the regime where only a handful of projections are available along an arbitrary scanning trajectory and folds the acquisition path itself into the inverse problem - a corner that operator-based toolboxes mostly leave untouched. From BAM and published in Review of Scientific Instruments 2020; run Main.py to launch. Niche and lightly starred, but there is little else that covers trajectories explicitly.
pyHST2 [site]	C / Python	See repository	-	The ESRF beamline reconstruction code — gridrec, FBP and iterative methods with phase retrieval for propagation-based phase-contrast tomography. Long-standing workhorse of the synchrotron community.
CTrex [site]	Python / CUDA (PyCUDA)	See source	-	A research-oriented GPU iterative reconstruction framework that exposes the projection–correction–backprojection pipeline all the way down to the individual CUDA kernels, so non-ideal acquisition effects can be folded into the operators themselves instead of being bolted on afterwards as corrections. Geometry is carried as per-projection homogeneous view matrices, which keeps the kernels agnostic to the scan path — circular, helical, offset and conveyor-belt trajectories all work, as do motion- and deformation-aware, event-based 4D and extended-field-of-view cases. Aimed squarely at micro-CT under unconventional acquisition. Described in arXiv:2609.30166; the paper announces no code repository, so this entry points at the preprint until one appears.

Toolkit

Language

License

Stars

Notes

TIGRE

MATLAB / Python

BSD-3-Clause

814

GPU-accelerated (CUDA) toolbox for cone-beam and parallel-beam CT; FDK, SART, SIRT, CGLS and TV variants.

ASTRA Toolbox [site]

C++ / Python / MATLAB

GPL-3.0

653

Long-standing research toolbox with a large set of projectors and algorithms; the reference CPU/GPU baseline.

TomoPy

Python

BSD-3-Clause

396

Reference toolkit for synchrotron micro/nano tomography; gridrec, FBP and phase retrieval.

RTK [site]

C++

Apache-2.0

298

Reconstruction Toolkit built on ITK; widely used in clinical CBCT and interventional research.

LEAP

CUDA / Python

MIT

257

LLNL's LivermorE AI Projector for Computed Tomography (LEAP); production-grade model-based iterative reconstruction for 3D transmission CT.

PyTomography

Python

MIT

178

End-to-end medical image reconstruction with quantification, including CT-based attenuation correction.

Core Imaging Library (CIL) [site]

Python

Apache-2.0

156

CCPi framework for variational CT reconstruction with a large catalogue of optimisation algorithms.

OMEGA

MATLAB / Python

GPL-3.0

111

Open-source multi-dimensional tomographic reconstruction software covering PET/SPECT/CT.

tomosipo

Python

GPL-3.0

107

Painless specification of 2D/3D projection geometries; wraps ASTRA and integrates with PyTorch.

DBT-Reconstruction

MATLAB

GPL-3.0

Open-source reconstruction toolbox for digital breast tomosynthesis, covering DBT-specific geometry alongside FBP and iterative solvers.

ToMoBAR [site]

Python

Apache-2.0 / GPL-3.0

Python library for model-based iterative reconstruction (FISTA, ADMM, ordered subsets) with the CCPi regularisers, built on ASTRA; developed at Diamond Light Source.

tomocupy [site]

Python

BSD-style (see repository)

GPU tomographic reconstruction from Argonne's Advanced Photon Source, built on CuPy for fast reconstruction of large synchrotron datasets.

svmbir

C / Python

BSD-3-Clause

Fast Super-Voxel Model Based Iterative Reconstruction with proximal-map plug-ins.

MBIRJAX [site]

Python

BSD-3-Clause

Model-based iterative reconstruction in JAX from Bouman's group, running on CPU or GPU; the successor to svmbir.

tofu

Python

LGPL-3.0

Helper layer on top of the UFO framework for fast GPU tomographic reconstruction pipelines.

libcbct

C++

CC BY-NC-SA 4.0

Lightweight cone-beam CT reconstruction library with FDK and iterative solvers; non-commercial licence.

mumott [site]

Python

GPL-3.0

Tensor-tomography reconstruction for reciprocal-space data, developed at DTU. Specialised rather than general-purpose, but it is the most complete open implementation of the tensor-tomography model.

DIRECTT Reconstruction GUI

Python

GPL-2.0

A Python GUI built on the ASTRA toolbox for reconstructing X-ray CT data, with the DIRECTT algorithm included. DIRECTT targets the regime where only a handful of projections are available along an arbitrary scanning trajectory and folds the acquisition path itself into the inverse problem - a corner that operator-based toolboxes mostly leave untouched. From BAM and published in Review of Scientific Instruments 2020; run Main.py to launch. Niche and lightly starred, but there is little else that covers trajectories explicitly.

pyHST2 [site]

C / Python

See repository

The ESRF beamline reconstruction code — gridrec, FBP and iterative methods with phase retrieval for propagation-based phase-contrast tomography. Long-standing workhorse of the synchrotron community.

CTrex [site]

Python / CUDA (PyCUDA)

See source

A research-oriented GPU iterative reconstruction framework that exposes the projection–correction–backprojection pipeline all the way down to the individual CUDA kernels, so non-ideal acquisition effects can be folded into the operators themselves instead of being bolted on afterwards as corrections. Geometry is carried as per-projection homogeneous view matrices, which keeps the kernels agnostic to the scan path — circular, helical, offset and conveyor-belt trajectories all work, as do motion- and deformation-aware, event-based 4D and extended-field-of-view cases. Aimed squarely at micro-CT under unconventional acquisition. Described in arXiv:2609.30166; the paper announces no code repository, so this entry points at the preprint until one appears.

Operators, solvers & classical algorithms (9)

Toolkit	Language	License	Stars	Notes
PyLops [site]	Python	LGPL-3.0	541	Linear-operator library with a Radon transform and a large set of solvers (CGLS, LSQR, TV, sparsity-promoting). The cleanest way to prototype a variational CT problem without writing your own matrix-free operators.
ODL [site]	Python	MPL-2.0	434	Operator Discretization Library — a functional-analytic framework for inverse problems, with CT geometries built in.
sigpy	Python	BSD-3-Clause	339	Signal-processing package for iterative image reconstruction; proximal operators and NUFFT primitives.
MIRT	Julia	MIT	126	Michigan Image Reconstruction Toolbox. Reference implementations of a very wide range of classical, statistical and penalised-likelihood reconstruction methods; the Julia port of the long-standing MATLAB MIRT.
leehoy/CTReconstruction	Python	GPL-2.0	97	Compact, readable Python and MATLAB CT reconstruction code. FBP plus ASD-POCS, SART and EM already exist under python/, though the README still lists iterative reconstruction as "to be added".
IRtools	MATLAB	See repository	91	Iterative Regularization Tools — hybrid Krylov methods, Tikhonov, TV and general-form regularisation, shipped with large-scale tomography test problems. From Hansen's group.
AIR Tools II	MATLAB	BSD-3-Clause	89	MATLAB package of algebraic iterative reconstruction methods (ART and SIRT families) with test problems, stopping rules and parameter choice, by Hansen and Jørgensen.
CCPi Regularisation Toolkit	C / Python	Apache-2.0	63	CPU/GPU implementations of TV, TGV, directional TV and other regularisers used by CIL.
pySART	Python	GPL-2.0	33	Compact reference implementation of the Simultaneous Algebraic Reconstruction Technique.

Toolkit

Language

License

Stars

Notes

PyLops [site]

Python

LGPL-3.0

541

Linear-operator library with a Radon transform and a large set of solvers (CGLS, LSQR, TV, sparsity-promoting). The cleanest way to prototype a variational CT problem without writing your own matrix-free operators.

ODL [site]

Python

MPL-2.0

434

Operator Discretization Library — a functional-analytic framework for inverse problems, with CT geometries built in.

sigpy

Python

BSD-3-Clause

339

Signal-processing package for iterative image reconstruction; proximal operators and NUFFT primitives.

MIRT

Julia

MIT

126

Michigan Image Reconstruction Toolbox. Reference implementations of a very wide range of classical, statistical and penalised-likelihood reconstruction methods; the Julia port of the long-standing MATLAB MIRT.

leehoy/CTReconstruction

Python

GPL-2.0

Compact, readable Python and MATLAB CT reconstruction code. FBP plus ASD-POCS, SART and EM already exist under python/, though the README still lists iterative reconstruction as "to be added".

IRtools

MATLAB

See repository

Iterative Regularization Tools — hybrid Krylov methods, Tikhonov, TV and general-form regularisation, shipped with large-scale tomography test problems. From Hansen's group.

AIR Tools II

MATLAB

BSD-3-Clause

MATLAB package of algebraic iterative reconstruction methods (ART and SIRT families) with test problems, stopping rules and parameter choice, by Hansen and Jørgensen.

CCPi Regularisation Toolkit

C / Python

Apache-2.0

CPU/GPU implementations of TV, TGV, directional TV and other regularisers used by CIL.

pySART

Python

GPL-2.0

Compact reference implementation of the Simultaneous Algebraic Reconstruction Technique.

Differentiable & learned reconstruction (11)

Toolkit	Language	License	Stars	Notes
DeepInverse [site]	Python	BSD-3-Clause	835	PyTorch library for imaging inverse problems — physics operators, learned denoisers, unrolled networks and benchmarks.
torch-radon	Python	GPL-3.0	281	Differentiable Radon transform and filtered backprojection in PyTorch.
MIRTorch	Python	See source	148	A PyTorch-powered differentiable image reconstruction and optimization toolbox for medical imaging.
PYRO-NN	Python / CUDA	Apache-2.0	121	Python Reconstruction Operators in Neural Networks (FAU Maier group): known-operator CT projectors as neural network layers for end-to-end learned reconstruction, with fan/cone-beam geometries.
Learned Primal-Dual	Python	See repository	110	Reference implementation of Learned Primal-Dual reconstruction — the canonical unrolled primal-dual scheme, originally applied to CT.
Learned Gradient Tomography	Python	See repository	96	Reference implementation of learned gradient descent for ill-posed tomography (Adler & Öktem); the origin of many later unrolled CT networks.
LION	Python	GPL-3.0	64	Learned Iterative Optimization Networks — unrolled reconstruction with PyTorch, used in CT and PET.
diffct [site]	Python / CUDA	Apache-2.0	28	Differentiable CUDA CT projectors from FAU (Maier group): arbitrary per-view trajectories, autograd through volume and acquisition geometry, exact-adjoint project/backproject pairs, multi-GPU/multi-node. Technical report 2026-10-14 (doi:10.20944/preprints202605.1446.v2).
PFITRE	Python	MIT	11	Perception Fused Iterative Tomography Reconstruction Engine — a CNN carrying perceptual knowledge as a regularizer inside an iterative solver, correcting limited-angle artefacts in X-ray nano-tomography. Ships a demo, a data-generation script and pretrained weights (Google Drive). From Brookhaven National Lab (NSLS-II); published in npj Computational Materials 2025.
X-Splat	Python	See source	5	Unified Gaussian splatting for joint RGB and X-ray 3D reconstruction: shared geometry across primitives, cross-modal registration via X-ray geometric proxies. Code of the SIGGRAPH Asia 2026 paper (also listed under Papers).
Dual-Domain-CT-Denoising	Python	See source	0	A deep learning framework for dual-domain low-dose CT denoising.

Toolkit

Language

License

Stars

Notes

DeepInverse [site]

Python

BSD-3-Clause

835

PyTorch library for imaging inverse problems — physics operators, learned denoisers, unrolled networks and benchmarks.

torch-radon

Python

GPL-3.0

281

Differentiable Radon transform and filtered backprojection in PyTorch.

MIRTorch

Python

See source

148

A PyTorch-powered differentiable image reconstruction and optimization toolbox for medical imaging.

PYRO-NN

Python / CUDA

Apache-2.0

121

Python Reconstruction Operators in Neural Networks (FAU Maier group): known-operator CT projectors as neural network layers for end-to-end learned reconstruction, with fan/cone-beam geometries.

Learned Primal-Dual

Python

See repository

110

Reference implementation of Learned Primal-Dual reconstruction — the canonical unrolled primal-dual scheme, originally applied to CT.

Learned Gradient Tomography

Python

See repository

Reference implementation of learned gradient descent for ill-posed tomography (Adler & Öktem); the origin of many later unrolled CT networks.

LION

Python

GPL-3.0

Learned Iterative Optimization Networks — unrolled reconstruction with PyTorch, used in CT and PET.

diffct [site]

Python / CUDA

Apache-2.0

Differentiable CUDA CT projectors from FAU (Maier group): arbitrary per-view trajectories, autograd through volume and acquisition geometry, exact-adjoint project/backproject pairs, multi-GPU/multi-node. Technical report 2026-10-14 (doi:10.20944/preprints202605.1446.v2).

PFITRE

Python

MIT

Perception Fused Iterative Tomography Reconstruction Engine — a CNN carrying perceptual knowledge as a regularizer inside an iterative solver, correcting limited-angle artefacts in X-ray nano-tomography. Ships a demo, a data-generation script and pretrained weights (Google Drive). From Brookhaven National Lab (NSLS-II); published in npj Computational Materials 2025.

X-Splat

Python

See source

Unified Gaussian splatting for joint RGB and X-ray 3D reconstruction: shared geometry across primitives, cross-modal registration via X-ray geometric proxies. Code of the SIGGRAPH Asia 2026 paper (also listed under Papers).

Dual-Domain-CT-Denoising

Python

See source

A deep learning framework for dual-domain low-dose CT denoising.

Simulation, phantoms & physics (9)

Toolkit	Language	License	Stars	Notes
xraylib [site]	C / Python	See repository	174	X-ray matter interaction cross sections, fluorescence yields and Compton profiles; the physics layer for realistic simulation.
GATE 10 (opengate) [site]	Python	LGPL-3.0	158	The Python generation of GATE, the Geant4-based Monte Carlo platform for medical imaging and radiotherapy; simulates CT acquisition, scatter and dose.
TomoPhantom	Python	Apache-2.0	134	Analytical 2D/3D/4D phantom generation plus forward projection for controlled reconstruction experiments. Also listed under Learning.
MC-GPU	CUDA	Public domain (FDA)	107	GPU Monte Carlo simulation of X-ray transport from the FDA, used to generate realistic CT and breast-imaging projections including scatter.
XCIST	Python	BSD-3-Clause	82	Open-source X-ray CT simulator with reconstruction, in the CatSim lineage; models geometry, spectra, detector response and physics, and generated the AAPM CT-MAR challenge data.
Physics-ArX	Python	See repository	79	Physics-based data-augmentation library that injects realistic CT/CBCT noise, scatter and artefact models into existing images to build paired training data.
LD-CT simulator	MATLAB	See repository	59	Low-dose CT simulator that converts high-dose images into realistic low-dose counterparts using a validated noise model.
XrayPhysics [site]	C++ / Python	MIT	56	X-ray cross sections, source-spectrum modelling, multi-material beam-hardening correction and dual-energy decomposition. From the LEAP author, and it pairs naturally with a simulator when you need a physically honest spectrum.
xtomosim	Python	See source	1	An X-ray computed tomography physical simulator in Python by Giavanna Jadick.

Toolkit

Language

License

Stars

Notes

xraylib [site]

C / Python

See repository

174

X-ray matter interaction cross sections, fluorescence yields and Compton profiles; the physics layer for realistic simulation.

GATE 10 (opengate) [site]

Python

LGPL-3.0

158

The Python generation of GATE, the Geant4-based Monte Carlo platform for medical imaging and radiotherapy; simulates CT acquisition, scatter and dose.

TomoPhantom

Python

Apache-2.0

134

Analytical 2D/3D/4D phantom generation plus forward projection for controlled reconstruction experiments. Also listed under Learning.

MC-GPU

CUDA

Public domain (FDA)

107

GPU Monte Carlo simulation of X-ray transport from the FDA, used to generate realistic CT and breast-imaging projections including scatter.

XCIST

Python

BSD-3-Clause

Open-source X-ray CT simulator with reconstruction, in the CatSim lineage; models geometry, spectra, detector response and physics, and generated the AAPM CT-MAR challenge data.

Physics-ArX

Python

See repository

Physics-based data-augmentation library that injects realistic CT/CBCT noise, scatter and artefact models into existing images to build paired training data.

LD-CT simulator

MATLAB

See repository

Low-dose CT simulator that converts high-dose images into realistic low-dose counterparts using a validated noise model.

XrayPhysics [site]

C++ / Python

MIT

X-ray cross sections, source-spectrum modelling, multi-material beam-hardening correction and dual-energy decomposition. From the LEAP author, and it pairs naturally with a simulator when you need a physically honest spectrum.

xtomosim

Python

See source

An X-ray computed tomography physical simulator in Python by Giavanna Jadick.

Artefact correction & calibration (4)

Toolkit	Language	License	Stars	Notes
Discorpy	Python	Apache-2.0	93	Sub-pixel camera and lens distortion calibration; used to correct detector/geometry distortions before reconstruction.
Algotom	Python	Apache-2.0	69	Data-processing algorithms for tomography — ring and artefact removal, distortion correction, denoising, phase retrieval and reconstruction. Developed for synchrotron beamlines.
Sarepy	Python	See repository	63	Numerical techniques for ring-artefact removal in tomography — the most complete open collection of stripe/ring filters.
CBCT Scatter Correction Tool	-	See repository	16	One of the very few open tools for cone-beam CT scatter estimation and correction; distributed as a prebuilt Windows executable, with no source in the repository.

Toolkit

Language

License

Stars

Notes

Discorpy

Python

Apache-2.0

Sub-pixel camera and lens distortion calibration; used to correct detector/geometry distortions before reconstruction.

Algotom

Python

Apache-2.0

Data-processing algorithms for tomography — ring and artefact removal, distortion correction, denoising, phase retrieval and reconstruction. Developed for synchrotron beamlines.

Sarepy

Python

See repository

Numerical techniques for ring-artefact removal in tomography — the most complete open collection of stripe/ring filters.

CBCT Scatter Correction Tool

See repository

One of the very few open tools for cone-beam CT scatter estimation and correction; distributed as a prebuilt Windows executable, with no source in the repository.

Data, viewers & platforms (7)

Toolkit	Language	License	Stars	Notes
MONAI [site]	Python	Apache-2.0	8797	PyTorch-based framework for medical imaging with CT reconstruction transforms and pretrained models.
3D Slicer [site]	C++	BSD-style (see repository)	2678	The dominant open platform for medical image computing; used to inspect, segment and register CT volumes and to prototype reconstruction pipelines as extensions.
pydicom [site]	Python	MIT	2225	Read, modify and write DICOM files — the standard entry point for clinical CT data.
tomviz [site]	C++	BSD-3-Clause	367	Cross-platform application for tomographic data visualisation and processing.
voxenra	Python	See repository	284	Cross-platform DICOM viewer for CT, MRI and PET/CT with multiplanar reformatting, 3D volume rendering and fusion — handy for eyeballing reconstruction output.
RadIO	Python	Apache-2.0	227	A CT preprocessing and augmentation framework oriented towards lung-nodule pipelines; despite the name, there is no Radon transform or FBP here.
CONRAD	Java	GPL-3.0	72	Java framework for cone-beam CT research from FAU Erlangen, covering simulation, phantoms and reconstruction, with GPU acceleration.

Toolkit

Language

License

Stars

Notes

MONAI [site]

Python

Apache-2.0

8797

PyTorch-based framework for medical imaging with CT reconstruction transforms and pretrained models.

3D Slicer [site]

C++

BSD-style (see repository)

2678

The dominant open platform for medical image computing; used to inspect, segment and register CT volumes and to prototype reconstruction pipelines as extensions.

pydicom [site]

Python

MIT

2225

Read, modify and write DICOM files — the standard entry point for clinical CT data.

tomviz [site]

C++

BSD-3-Clause

367

Cross-platform application for tomographic data visualisation and processing.

voxenra

Python

See repository

284

Cross-platform DICOM viewer for CT, MRI and PET/CT with multiplanar reformatting, 3D volume rendering and fusion — handy for eyeballing reconstruction output.

RadIO

Python

Apache-2.0

227

A CT preprocessing and augmentation framework oriented towards lung-nodule pipelines; despite the name, there is no Radon transform or FBP here.

CONRAD

Java

GPL-3.0

Java framework for cone-beam CT research from FAU Erlangen, covering simulation, phantoms and reconstruction, with GPU acceleration.