TIGRE#

The Tomographic Iterative GPU-based Reconstruction toolbox: fast, flexible 3D CT reconstruction for MATLAB and Python.

TIGRE reconstructs tomographic images from any scanning geometry, using GPU implementations (including multi-GPU) of the forward and backprojection operators and a wide range of iterative algorithms.

Install TIGRE

View on GitHub

Two reconstructions of a coffee bean slice, side by side

Micro-CT of a coffee bean, 3340 × 3340 × 900 voxels from 2134 projections, reconstructed with FDK and with CGLS.#

Why TIGRE#

Any geometry

Cone beam, parallel beam, helical, tomosynthesis, C-arm and curved detectors. Geometric parameters are defined per projection, not per scan.

Fast on GPUs

CUDA projectors and backprojectors that split large problems across multiple GPUs automatically.

Many algorithms

FDK/FBP, SART-family, Krylov subspace methods (CGLS, LSQR, LSMR, GMRES), MLEM, total-variation regularised methods and more.

MATLAB and Python

The same toolbox and the same demos in both languages.

Real data

Loaders for Nikon, Varian, Philips (DICOM) and other scanner formats.

Open source

BSD-licensed, developed at the University of Bath and CERN with contributions from the community.

Get started#

Install

MATLAB and Python installation.

Install
User guide

Geometry, projectors, algorithms and more.

User guide
Demos

Worked examples in Python and MATLAB.

Demos
Cite TIGRE

How to cite TIGRE in your work.

Citing TIGRE