About the project
In this industry sponsored project, you will develop advanced machine learning tools to detect and remove scatter artefacts from X-ray images acquired with state of the art, high energy X-ray imaging systems, where scatter becomes the dominant source of image artefact.
X-rays are used extensively to visualise the inside of objects. Not only are they one of the key modalities used in medical imaging, they are also becoming increasingly important in industrial applications where they are a key technology for the non-destructive inspection of complex components and assemblies.
X-ray based computed tomography (XCT) is a particularly valuable engineering tool that can provide high resolution and magnified 3D representations of internal object geometries, quantitative information and defect mapping. XCT enables the inspection of complex components and assemblies and thus guarantees safety, accuracy and/or efficiency in virtually every major industry.
Unfortunately, using commonly available X-ray tomography systems, object size currently remains a major limiting factor, particularly for objects made of materials such as metals, which are relatively opaque to X-rays. Standard commercial X-ray tomography systems generate X-rays with energies in the kilo electron volt (keV) range, which significantly limits the amount of material they can penetrate.
An alternative is to use more powerful X-ray sources that generate X-rays with energies in the MeV region. Whilst MeV linear accelerator sources are now becoming more generally available for industrial inspection and whilst experimental laser based X-ray sources are being build in dedicated scientific research labs, a range of challenges remain preventing their routine use. This is partly due to a change in the relevant physical principles that govern the interactions between high energy (MeV) X-rays and matter. X-ray images acquired at these energies become increasingly contaminated by scattered X-ray radiation, which obscures and distorts relevant information.
This project is one of two related, industry sponsored PhDs that will look at the reduction and removal of the scatter signal from X-ray tomographic data acquired with MeV sources. The novel aspect of this project will be the investigation of advanced machine learning methods to reduce and remove the effect of X-ray scattering from high energy X-ray tomographic data.
These computational methods predict, and thus remove, the contribution of the scattered X-ray radiation. Two approaches are feasible:
- predicting the scatter signal directly from the observed data
- using the knowledge of the forward propagation of X-rays through materials, coupled with estimated or know object geometries, to simulate (and then remove) the X-ray scatter signal
Using simulated and real X-ray images, this project will study and modify these approaches and study how they perform when applied to high energy (>MeV) X-ray tomography data.
You will be working alongside the team at the University of Southampton’s µ-VIS facility, a dedicated X-ray Computed Tomography (XCT) centre. The centre is part of the UK’s National Facility for lab-based XCT and houses some of the UK’s largest micro-focus CT scanning systems, capable of unveiling sub-surface information from materials, components and structures. With strong links between both research and industry, the centre itself is used for aircraft crash investigations, Formula 1, space technology and much more.
