X-ray scanners can now see not just the shape of an object, but what it is chemically made of, in five dimensions. This "colour X-ray" technology adds a chemical dimension to standard 3D and 4D imaging, but it remains locked away from most researchers because the data it produces is enormous and difficult to process, and the images are slow and low-resolution. This project will build new imaging systems and create accessible software workflows—using machine learning and sensor fusion—to make the technology fast, high-resolution, and usable by non-specialists. If successful, the work could transform non-destructive testing across multiple industries. In medicine, it could enable multiple staining methods for clearer biological imaging. In manufacturing and recycling, it could map rare earth elements inside magnets or heavy metals in electronic waste without destroying the material. The team will also study chemical reactions in real time inside batteries and catalysts. The project is applied and hardware-focused, with clear routes to industrial and clinical use.
View original technical description
The next dimension of X-ray imaging is here! Five dimensional Colour X-ray Imaging enables the non-destructive 3D and 4D images to be complemented by a fifth chemistry dimension. Whilst the fundamental technology has been demonstrated it is, at present, inaccessible to most researchers due to the difficulty of acquisition and the processing of the vast multidimensional datasets it generates. In addition, slow speed acquisition of limited resolution images is a barrier to its utility. This project will build new Colour X-ray Imaging systems, create accessible workflows, including new reconstruction algorithms, machine learning analysis and sensor fusion, which will not only make the technology usable for a wide range of users but will overcome barriers in resolution and speed of detection. We will also be pushing forward our understanding on the detection limits and quantification of the compositional data recorded which is of crucial importance for greater uptake. We will combine hardware developments through the complementary use of hyperspectral, multispectral and conventional X-ray detectors. High speed, high resolution data will be collected via sensor fusion achieved through acquisition procedures through optimisation of the different detector types. Efficient and effective analysis of the data will be enabled through machine learning approaches to identify the relevant signals that demand reconstruction into a virtual 5D replica of the sample which include spatial, temporal and chemical dimensions. Lastly, the high-speed data acquisition with spectral information through data fusion will enable us to study chemical reactions in-situ, notably in the field of catalysis and batteries, while super-resolution image fusion will allow for unprecedented non-destructive 3D mapping of elements in natural minerals and synthetic materials, e.g. for the recycling of heavy metals from slags and electronic waste and rare earth elements from magnets. Merging multispectral and hyperspectral imaging will enable to push forward the development of staining materials and the applications of multiple staining methods for (bio)medical imaging.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know