X-ray micro-computed tomography (micro-CT) is a powerful imaging technique used across industries such as energy, automotive, and advanced manufacturing to examine the internal structure of materials and devices. However, imaging certain components---particularly those containing metals---presents significant challenges due to beam hardening. This effect occurs when low-energy X-rays are disproportionately absorbed by dense materials, leading to artefacts that obscure critical details. In extreme cases, photon starvation can completely mask regions of interest, reducing the reliability of the final reconstructions.
To address these issues, this project will develop advanced computational techniques for improving the quality of micro-CT images acquired from multi-orientational scans. A common approach to mitigating beam hardening is to capture multiple scans of the sample at different orientations and combine them to enhance the reconstructed image. However, this process is currently manual, time-consuming, and prone to errors. The goal of this project is to create a more efficient and semi-automated pipeline for processing these datasets.
The key innovations in this project include:
* Improved Image Combination Strategies -- In collaboration with NPL (National Physical Laboratory), we will investigate methods for optimally integrating data from multiple orientations, reducing beam hardening effects while preserving fine structural details.
* (Semi-)Automated Registration -- We will develop a method using physical reference markers embedded in the sample, such as small metallic spheres, to ensure accurate alignment of multi-orientational scans. This approach will improve reproducibility and reduce the need for manual adjustments.
* Self-Supervised Denoising -- We will apply cutting-edge self-supervised machine learning techniques to enhance the signal quality of micro-CT reconstructions without requiring extensive training datasets. This will allow for improved image clarity while maintaining computational efficiency.
* Advanced Segmentation Tools -- The project will focus on automatically distinguishing different material components in reconstructed volumes. A specific application will be the segmentation of copper (Cu) tracing in printed circuit boards (PCBs), a challenge that is highly relevant to industries dealing with electronics inspection, where beam hardening significantly affects image quality.
By improving the accuracy and efficiency of micro-CT data processing, this project will enable faster and more reliable analysis of complex materials and electronic components. The outcomes will support industries that rely on high-quality imaging for quality control, failure analysis, and product development, ultimately advancing the capabilities of non-destructive testing techniques.
Finden and NPL are working together to develop novel statistical and machine-learning based methods to reduce the noise levels for chemical imaging and tomography datasets. The proposed approaches will result in clearer images with sharper associated spectra/patterns, aiding interpretation and quantification of the data. The approach will be applicable to many different forms of hyperspectral/scattering based characterisation, both chemical mapping and chemical computed tomography (CT) e.g. XRF-CT/mapping, XRD-CT/mapping, IR , as well as having potential benefits to more conventional imaging methods such as X-ray-CT.
Successful denoising will allow us to work with weaker signals than before, opening up possibilities for faster measurement times (resulting in cost savings that can be passed on to the customer), resulting in higher throughput of chemical characterisation, but also the option of maintaining image quality but with a reduced X-ray dose - of benefit to medical imaging.
Through this project we seek to identify and develop methods to automatically segment complex chemical imaging datasets, e.g. XRD-CT, Raman mapping and other hyperspectral imaging techniques. The goal is to identify the minimum number of unique chemical environments in a dataset, without needing prior knowledge or input as to the expected identity or number of components present.
The output from this segmentation will then be used to inform subsequent data quantification and analysis steps, and also to see if it is possible to identify correlations between each of these components. A successful outcome will be of benefit to a broad range of industry and chemical services companies, as many analytical methods suffer from similar segmentation challenges.
Our company has developed advanced chemical imaging capabilities which we offer as a service to industry, helping our clients accelerate their R&D. Our imaging approaches yield rich and large datasets that contain an abundance of physico-chemical information. This project will use artificial intelligence approaches to reconstruct X-ray scatter-based chemical tomography data from large objects.Large objects pose a problem due to geometric blurring of the scattered signals on the receiving detector, preventing conventional reconstruction approaches. We have spent considerable resources developing a non-linear least-squares algorithm to address this but it is computationally demanding and because of this imposes resolution limits on the reconstructed data (i.e. small images size). We have realised though that the problem has several features which indicate that it can be tackled by using deep learning approaches. Additionally, we have the ability to generate very large simulated labelled datasets that can be used as training sets for supervised learning using convolutional neural networks (CNNs). This is in addition the very large real data sets we have at our disposal. Whilst there are existing attempts to reconstruct conventional tomography data using CNNs, we are planning to develop new CNNs for reconstructing chemical (hyperspectral) tomography data and indeed overcome the parallax problem. The project thus is innovative both in terms of approach and application and will push the opportunities in this emerging field.
This project will use machine learning approaches to extract physico-chemical information from chemical imaging data. This novel approach will tackle an emerging problem in this field, namely how to automatically identify and extract chemical signals from the rich and ever-larger datasets that it is now possible to collect. There are several features that suggest this problem can be tackled using machine learning approaches. We have developed software for the rapid simulation of chemical imaging data, and we can use this to generate large labelled datasets for training the convolutional neural networks (CNN) that we will build. In addition we have substantial libraries of real data which the developed CNN's can be tested against.
"Air pollution is one of today's most concerning problems, causing unacceptable health issues to many people in our societies. This project will strive to deliver materials, and products for easy customer use, that can help tackle this problem for UK and global benefit.
The project will take novel technology now emerging from the UK's academic (University College London) and business (Johnson Matthey) communities and optimise its use in a widely used product such as decorative paint using cutting-edge chemical imaging technology (Finden).
This project will deliver high quality decorative paints (Dulux from Imperial Chemical Industries) that offer the customer the additional functionality of air depollution in their homes, their schools and their places of work to compliment the colour and protection that Dulux paints currently offer. This will focus especially on tackling indoor air pollution from carbon monoxide, benzene, nitrogen oxides and formaldehyde.
Such functional decorative paints will also act as a springboard for uptake of this novel technology in other polymer-based coatings, such as active food packaging and respiratory protection devices, and the project will also look to exploit new chemical imaging technology."