Emanuel Larsson illustrates how the latest version of his kitchen-based light tomography (KBLT) scanner works. Photo: Noomi Egan.

A late Friday night in December 2018, Emanuel Larsson, who was working at RISE Research Institutes of Sweden at the time, started to assemble a small-scale tomography scanner in his kitchen, made up by everyday objects such as a camera, a water bottle, lego pieces, and a flash light. Why? To be able to better teach tomographic imaging to Swedish industry representatives and students. Today, his model is used as a prototype to help prepare for future neutron beamlines at ESS - as part of a collaborative project between LTH and ESS, hosted by LINXS this summer.

– I had noticed that it was quite hard for people to grasp the basic principle of tomographic imaging. That made me want to build a model where you can actually show what is going on. I used stuff I found in the house, and I realised that they worked well to illustrate how one can scan, reconstruct and render a sample in 3D, says Emanuel Larsson.

Neutron tomography is an imaging technique that allows visualisation of samples in three dimensions. The sample rotates in the beam and multiple 2D radiography images are recorded with high-speed digital cameras. A 3D representation of the volume of the object can be reconstructed using a mathematical algorithm. The technique also enables the visualisation of fluids, such as water or oil in large metal objects.

Neutron tomography is pivotal to many future breakthroughs since it is non-destructive and can thus be used to investigate the temporal and spatial resolution down to the micrometer scale in samples of both soft and hard matter- for example energy- and engineering materials, biological and geological samples and cultural heritage artefacts.

The model has evolved over time

The very first version of the KBLT scanner was very basic and built of objects such as a paper screen, a light, a camera and a sample such as a water bottle. Credit: Emanuel Larsson.

Emanuel Larsson’s first model was very basic. He used a paper screen, a light, a camera and a sample. He took a photo of the sample, for example a water bottle, then he rotated it and took a new image, until he had captured the sample over 360 degrees. Three years later, that very first model has evolved – and now Emanuel is on version 25. The new model is refined but is still using the same flash light, and is now connected to a single board Raspberry Pi 4 computer. The sample can rotate 200 steps over 360 degrees, which should be compared to tomography setups at large scale research facilities were often thousands of radiographic images are captured of a sample over 360 degrees. In addition, Emanuel has also set up user-friendly image reconstruction and analysis pipeline which can be applied both on the acquired and reconstructed data sets.

– Even though the model is quite basic it does the job. We did a comparison where we scanned a legoman with my model, and with real x-rays with the lab tomograph in the 4D Imaging Lab at the Division of Solid Mechanics at Lund University. And the results were not that different! The regular light could of course not penetrate through the plastic, as with the x-rays, but in both experiments you can see the outline of the figure in 3D.

The model is used as a prototype for ODIN at ESS

Stefanos Athanasopoulos, Samuel Staines and Emanuel Larsson have collaborated on the project which aims to help prepare for future neutron beamlines at ESS, hosted at LINXS this summer. Photo: Noomi Egan. — Stefanos Athanasopoulos, Samuel Staines and Emanuel Larsson have collaborated on the project which aims to help prepare for future neutron beamlines at ESS, hosted at LINXS this summer. Photo: Noomi Egan.

Today, Emanuel’s model is not only used in the teaching he does in his new positions as a Method Expert in X-ray and Neutron Imaging at the Division of Solid Mechanics and as an Application Expert in Tomographic Imaging and Image Analysis at LUNARC, the Center for Scientific and Technical Computing at Lund University, it is also used as a prototype for the test beamline YMIR for the future imaging beamline ODIN, which will be built at ESS. This is because the techniques employed in the model can be used to test both hardware and software pipelines both with light tomography and neutron tomography – even though Emanuel’s model fits on two tables, and a neutron beamline fits in a very large room!

Samuel Staines, a master student at the Division of Solid Mechanics, is responsible for designing a replica of the model which can be compatible with what is needed for the new YMIR test beamline and the future ODIN beamline at ESS. He explains that it works just as well to use normal light as it would to use neutrons in order to test out how the beamline might behave. Before the ODIN beamline is built, the techniques will be tested in the test beamline, YMIR.

– There is a huge amount of work you need to do to develop a new beamline. That is why you need to test the platform, including hardware and software pipelines in multiple ways. One way is to refine and test in a smaller scale to get ideas on where the challenges and pitfalls will lie with the new instrument, says Samuel Staines.

Improving data collection strategies

Stefanos Athanasopoulos, who is a Guest Researcher at the Division of Solid Mechanics is also involved in the project with ESS. His work is focused on improving the data collection strategies for future experiments at large scale research facilities.

– A big part of developing a new beamline is also about testing optimal ways for data collection. How will it actually work when a user comes in and runs an experiment? At the moment a lot of data is acquired from the experiments, says Stefanos Athanasopoulos.

– The ideal solution would be to build a single log file with all the data and information about the experimental setup so that you get everything in one place. That file can then be used by different researchers, not just the person conducting the actual experiments.

The model makes techniques available to more students

Emanuel Larsson is also planning to make the code of his model freely available to students across the world – with the help of LINXS and Stephen Hall, outgoing Director of LINXS, and Senior Lecturer at the Division of Solid Mechanics at LTH. The idea is to publish an article on the kitchen tomography lab scanner, along with making the code available to download from the platform GitHub. If this plan succeeds, the cost of teaching students tomography techniques can be reduced drastically. Today, x-ray tomography scanners in laboratories can cost between 40 k€ to 1 M€. In comparison, the kitchen tomography scanner model can be built with readily available materials for between 100€ to 1000€, depending on which version of the KBLT scanner the students would like to build.

– Hopefully, the model can help inspire more students to start using imaging techniques in their research early on. It allows them to play around with samples and sample environments – for example what will happen if you point a heat gun onto a rotating sample, and image the change in 3D over time, thus meaning 4D?

While using the model, students will also learn important skills in engineering, image reconstruction, analysis and 3D-rendering. Skills that will come in handy for future imaging experiments both at ESS and MAX IV.

– In a few years models like the kitchen tomography scanner could have huge impact in terms of increasing the user base of people using the techniques offered at large scale research facilities. It is all about inspiring people and making it easy to learn, test things and have fun at the same time, Emanuel Larsson concludes.