Overview¶
Mantid Imaging contains tools for loading data from neutron imaging experiments, preprocessing to enhance images and tomographic reconstruction into 3 dimensional data that can be output for further analysis.
Mantid Imaging makes use of a range of algorithms, some from external tools including Astra, Tomopy, Algotom and the Core Imaging Library. Many of these are optimised for multithreading and GPU computing.
Mantid Imaging is written in Python and requires a CUDA capable GPU for full functionality. It runs on Linux. See Installation for more details.
See the User Guide for an example taking data from a set of images to full reconstruction.
Features¶
Operations¶
The Operations Window provides a selection of tools and filters to process and enhance the 2D images before running the reconstruction.
Crop Coordinates
Flat-fielding
Remove Outliers
ROI Normalisation
Arithmetic
Circular Mask
Clip Values
Divide
Gaussian
Median
Monitor Normalisation
NaN Removal
Rebin
Rescale
Ring Removal
Rotate Stack
Remove all stripes
Remove dead stripes
Remove large stripes
Stripe Removal
Remove stripes with filtering
Remove stripes with sorting and fitting
More details on using these are available in the API Reference and User Guide.
Reconstruction¶
Mantid Imaging offers several reconstruction algorithms
FBP_CUDA - Filtered Backprojection
SIRT_CUDA - Simultaneous Iterative Reconstruction Technique
CIL: PDHG-TV - Core Imaging Library
gridrec
Glossary¶
- COR
Centre of Rotation.
- Dataset
Consists of images stacks plus additional files or metadata, such as flat and dark frames, and log files.
- Image
A single 2D recorded by the experiment. Typically loaded from TIFF files.
- Image Stack
A set of multiple related images, for example images of the same object from a range of angles.
- Projection
The original image formed by the unscattered/unabsorbed neutrons reaching the imaging detector.
- ROI
Region of Interest. A cropped part of an image.
- Sinogram
A projection of all rotations of a single slice through an object. Each point in the slice becomes a sinusoidal line. Also known as a Radon transform.
- Tomography
The process of creating a 3D model if an object from a series 2D images taken at a range angles through it.