Open X-ray Tomographic Datasets

This page contains following datasets:

STEMPO - dynamic X-ray tomography phantom dataset
3D cone-beam computed tomography dataset of a chicken bone imaged at 4 different dose levels
3D cone-beam computed tomography dataset of a walnut
3D cone-beam computed tomography dataset of a pine cone
3D cone-beam computed tomography dataset of a seashell
Tomographic X-ray data of a lotus root slice filled with different chemical elements
Tomographic X-ray data of a walnut
Tomographic X-ray data of a carved cheese
Tomographic X-ray data of 3D emoji
Tomographic X-ray data of 3D cross
Gel phantom data for dynamic X-ray tomography
Cone-beam Gel phantom data for dynamic X-ray tomography

STEMPO - dynamic X-ray tomography phantom dataset

Tommi Heikkilä

September 23, 2022

Photography of the STEMPO device and different views (xy-, xz- and yz-planes) of the volume changing over time. Reconstructed with a variational method similar to the included 2D example.

4D recon of STEMPO — Photography of the STEMPO device and different views (xy-, xz- and yz-planes) of the volume changing over time. Reconstructed with a variational method similar to the included 2D example.

The Spatio-TEmporal Motor-POwered (STEMPO) phantom is a physical phantom designed for collecting dynamic X-ray tomography data. The dynamic part of the phantom is computer controlled allowing for wide variety of different measurements and sampling setups to be used.

This data set can be appended with new data in the future. Current version (1.2.0) includes

Static scan (2D and 3D geometry)
Dynamic scan with 360 projections covering one rotation (2D and 3D geometry)
Dynamic scan with 8x45 scans covering 8 rotations (2D and 3D geometry)
Dynamic scan with 8x180 scans covering 8 rotations (2D and 3D geometry)
Approximate ground truth obtained via interpolation (2D only)
Example codes using FBP (filtered back-projection) and FDK (Feldkamp-Davis-Kress)
Example code using variational methods and wavelet regularization on the 2D + time object.
Example code using low-rank + sparse decompositon scheme and 2D wavelets.

The data is licensed under an open data license (CC BY 4.0) and it is hosted on Zenodo.

The data format is designed to be compatible with HelTomo Toolbox and ASTRA Toolbox.

Dataset:

https://doi.org/10.5281/zenodo.7081688

Detailed documentation:

arXiv:2209.12471 (preprint)

3D cone-beam computed tomography dataset of a chicken bone imaged at 4 different dose levels

Alexander Meaney

August 16, 2022

This is an open-access dataset of a series of 3D cone-beam computed tomography scans (CBCT) of a chicken bone imaged at 4 different dose levels. The dose was changed by varying the X-ray tube current and detector exposure time appropriately. The dataset consists of 4 scans, each containing

the projection images from the scan, with background corrections and flat-field corrections already applied,
a metadata file describing the measurement geometry and settings,

The data is licensed under an open data license (Creative Commons Attribution 4.0 International) and it is hosted on Zenodo.

The data format is designed to be compatible with HelTomo, the Helsinki Tomography Toolbox, available on Github.

Dataset:

https://zenodo.org/record/6990764

3D cone-beam computed tomography dataset of a walnut

Alexander Meaney

August 16, 2022

A collage of a photograph and CBCT images of the walnut — L-R: a log-transformed intensity projection, slices along the xy-, xz-, and yz- planes from an FDK reconstruction, a 3D rendering of the FDK reconstruction.

This is an open-access dataset of a 3D cone-beam computed tomography scan (CBCT) of a walnut. The dataset consists of

the projection images from the scan, with background corrections and flat-field corrections already applied,
a metadata file describing the measurement geometry and settings,
pictures of an example reconstruction.

The data is licensed under an open data license (Creative Commons Attribution 4.0 International) and it is hosted on Zenodo.

The data format is designed to be compatible with HelTomo, the Helsinki Tomography Toolbox, available on Github.

Dataset:

https://zenodo.org/record/6986012

3D cone-beam computed tomography dataset of a pine cone

Alexander Meaney

August 16, 2022

A collage of a photograph and CBCT images of the pine cone — L-R: a photograph of the sample, a log-transformed intensity projection, slices along the xy-, xz-, and yz- planes from an FDK reconstruction, a 3D rendering of the FDK reconstruction.

This is an open-access dataset of a 3D cone-beam computed tomography scan (CBCT) of a pine cone. The dataset consists of

the projection images from the scan, with background corrections and flat-field corrections already applied,
a metadata file describing the measurement geometry and settings,
pictures of the sample and an example reconstruction thereof.

The data is licensed under an open data license (Creative Commons Attribution 4.0 International) and it is hosted on Zenodo.

The data format is designed to be compatible with HelTomo, the Helsinki Tomography Toolbox, available on Github.

Dataset:

https://zenodo.org/record/6985407

3D cone-beam computed tomography dataset of a seashell

Emma Kamutta, Sofia Mäkinen, and Alexander Meaney

August 11, 2022

A collage of a photograph and CBCT images of the seashell — L-R: a photograph of the sample, a log-transformed intensity projection, slices along the xy-, xz-, and yz- planes from an FDK reconstruction, a 3D rendering of the FDK reconstruction.

This is an open-access dataset of a 3D cone-beam computed tomography scan (CBCT) of a seashell. The dataset consists of

the projection images from the scan, with background corrections and flat-field corrections already applied,
a metadata file describing the measurement geometry and settings,
pictures ofthe measurement setup, the sample, and an example reconstruction thereof.

The data is licensed under an open data license (Creative Commons Attribution 4.0 International) and it is hosted on Zenodo.

The data format is designed to be compatible with HelTomo, the Helsinki Tomography Toolbox, available on Github.

Dataset:

https://zenodo.org/record/6983008

Tomographic X-ray data of a lotus root slice filled with different chemical elements

Tatiana A. Bubba, Andreas Hauptmann, Simo Huotari, Juho Rimpeläinen and Samuli Siltanen

A picture of two reconstructions from the data by two different methods. — Reconstructions from data by MATLAB routines Lotus_Landweber.m (120 projection angles, on the left) and Lotus_Tikhonov.m (20 projection angles).

This is an open-access dataset of tomographic X-ray data of a slice of a lotus root. The dataset consists of

the X-ray sinogram of a single 2D slice of the lotus root with two different resolutions, and
the corresponding measurement matrices modeling the linear operation of the X-ray transform.

Also MATLAB code for reconstructions using filtered back-projection, Landweber iteration, and Tikhonov regularization are provided.

The data is licensed under an open data license (CC BY 4.0) and it is hosted on Zenodo.

An informal video report of the data measurement session is available hera as a video:

Documentation at arXiv:

https://arxiv.org/abs/1609.07299

Dataset:

https://doi.org/10.5281/zenodo.1254204

Tomographic X-ray data of a walnut

Keijo Hämäläinen, Lauri Harhanen, Aki Kallonen, Antti Kujanpää, Esa Niemi and Samuli Siltanen

A picture of three reconstructions from the data using different numbers of projections. — "Ground truth" reconstruction computed from 1200 projections using FBP (left), Tikhonov reconstruction from 120 projections (middle), and Tikhonov reconstruction from 20 projections.

This is an open-access dataset of tomographic X-ray data of a walnut. The dataset consists of the X-ray sinogram of a single 2D slice of the walnut with three different resolutions and the corresponding measurement matrices modeling the linear operation of the X-ray transform. Each of these sinograms was obtained from a measured 120-projection fan-beam sinogram by down-sampling and taking logarithms. The original (measured) sinogram is also provided in its original form and resolution.

In addition, a larger set of 1200 projections of the same walnut was measured and a high-resolution filtered back-projection reconstruction was computed from this data; both the sinogram and the FBP reconstruction are included in the dataset, the latter serving as a ground truth reconstruction.

The data is licensed under an open data license (CC BY 4.0) and it is hosted on Zenodo.

Documentation at arXiv:

https://arxiv.org/abs/1502.04064

Dataset:

https://doi.org/10.5281/zenodo.1254206

Tomographic X-ray data of a carved cheese

Tatiana A. Bubba, Markus Juvonen, Jonatan Lehtonen, Maximilian März, Alexander Meaney, Zenith Purisha and Samuli Siltanen

Image of a carved cheese and two of its reconstructions using 45 and 15 projections. — Image of a carved cheese and its reconstructions from 45 projections (middle) and 15 projections.

This is an open-access dataset of tomographic X-ray data of a carved cheese. The dataset consists of

the X-ray sinogram of a single 2D slice of the cheese slice with three different resolutions, and
the corresponding measurement matrices modeling the linear operation of the X-ray transform.

Each of the sinograms was obtained from a measured 360-projection fan-beam sinogram by down-sampling and taking logarithms. The original (measured) sinogram is also provided in its original form and resolution.

The data is licensed under an open data license (CC BY 4.0) and it is hosted on Zenodo.

An informal video report of the data measurement session is available here as a video:

Documentation at arXiv:

https://arxiv.org/abs/1705.05732

Dataset:

https://doi.org/10.5281/zenodo.1254210

Tomographic X-ray data of a time-dependent emoji phantom

Alexander Meaney, Zenith Purisha and Samuli Siltanen

Pictures of emoji and their reconstructions from X-ray data.

This is an open access dataset of tomographic X-ray data of emoji phantom. The dataset consists of

the X-ray sinogram of a single 2D slice of 33 emoji faces (contains 15 different emoji faces) made by small squared ceramic stones, and
the corresponding static and dynamic measurement matrices modeling the linear operation of the X-ray transform.

Each of these sinograms was obtained from a measured 60-projection fan-beam sinogram by down-sampling and taking logarithms. The original (measured) sinogram is also provided in its original form and resolution.

The data is licenced under an open data licence (CC BY 4.0) and it is hosted on Zenodo.

Documentation at arXiv:

https://arxiv.org/abs/1802.09397

Dataset:

https://doi.org/10.5281/zenodo.1183532

Tomographic X-ray data of time-dependent 3D cross phantom

Salla Latva-Äijö, Alexander Meaney and Samuli Siltanen

Picture of a cross and its reconstruction. — The phantom in the preparation phase (left), the ready-made phantom on the sample holder (middle), and two examples of the resulting 2D projection images (right).

This is an open access dataset of measured tomographic X-ray data of a 3D cross phantom. The dataset consists of

the X-ray sinogram with 16 or 30 time frames (depending on the resolution) of 2D slices of the phantom, and
the static and dynamic measurement matrices modelling the linear operation of the X-ray transform.

Each of these sinograms was obtained from a measured 360-projection fan-beam sinogram by down-sampling and taking logarithms. The original (measured) sinogram is also provided in its original form and resolution.

The data is licenced under an open data licence (CC BY 4.0) and it is hosted on Zenodo.

Documentation at arXiv:

https://arxiv.org/abs/1809.00166

Dataset:

https://doi.org/10.5281/zenodo.1446516

Gel phantom data for dynamic X-ray tomography

Tommi Heikkilä, Hanna Help and Alexander Meaney

March 5, 2020

Reconstructions of a gel phantom using Tikhonov regularization and FBP — Reconstructions from selected time steps. Top row: Tikhonov regularization, 30 projections. Bottom row: FBP, 360 projections.

This is an open-access dataset of tomographic X-ray data of a dynamic agarose-gel phantom perfused with a liquid contrast agent. The dataset consists of 17 consecutive X-ray measurements of a 2D slice of the gel phantom. This data is provided in two resolutions and stored in a special CT-data structure containing the metadata of the measurement setup.

In addition a denser projection sampling of the first time step, and a very dense sampling of an addition 18th time step is provided as a complementary data.

The data does not contain the corresponding forward operators as is, but allows for easy creation of computationally efficient matrix-free operators and FBP reconstructions, both on CPUs and on CUDA cabable GPUs. The use of said operators is also demonstrated with short example codes.

The data is licenced under an open data licence (CC BY 4.0) and it is hosted on Zenodo.

Documentation at arXiv:

https://arxiv.org/abs/2003.02841

Dataset:

https://doi.org/10.5281/zenodo.3696816

Cone-beam Gel phantom data for dynamic X-ray tomography

Tommi Heikkilä, Hanna Help and Alexander Meaney

April 28, 2023

FDK Reconstruction of the gel phantom at time step 12. Iodine contrast agent is shown in red.

This is an open-access dataset of tomographic cone-beam X-ray data of a dynamic agarose-gel phantom perfused with a liquid contrast agent. The dataset consists of 17 consecutive X-ray measurements of a 3D volume of the gel phantom. This data is provided in three resolutions and stored in a special CT-data structure containing the metadata of the measurement setup.

In addition a denser projection sampling of the first time step, and a very dense sampling of an addition 18th time step is provided as a complementary data.

The data does not contain the corresponding forward operators as is, but allows for easy creation of computationally efficient matrix-free operators and FDK reconstructions on CUDA cabable GPUs.

The data is licenced under an open data licence (CC BY 4.0) and it is hosted on Zenodo.

Dataset and documentation:

https://doi.org/10.5281/zenodo.7739984