Advanced X-Ray Micro-CT Methods for The Delineation of Tumour Heterogeneity in Lung Cancer

1) Brief description of the context of the research including potential impact

Analysis of tumour texture heterogeneity using machine learning techniques with CT-images shows potential for the prediction of cancer malignancy and aggressiveness [1], which may inform patient treatment plans. Phase-contrast x-ray techniques, which derive image contrast from phase-shifts in propagating x-rays [2], provide improved sensitivity to texture and microstructure in samples that would appear mostly homogenous to conventional, attenuation-based x-ray imaging.

This project proposes to investigate x-ray phase-contrast micro-CT as a method to image ex-vivo lung samples and produce three dimensional representations. The new approach will be compared to the current histopathological standard, which will be the ground truth. Additionally, machine learning and other advanced computational techniques will be applied for the registration, segmentation, and classification of the images.

This work will contribute to the efforts in scaling up these techniques towards new in-vivo clinical applications. 2) Aims and Objectives

The project will entail a bottom-up approach, starting with the modelling, optimisation and fine tuning of the imaging setups. Image retrieval algorithm development will follow, along with an imaging protocol for a pilot study with clinical tissue samples. From both modelling and the pilot data on ex-vivo tissue samples, it will be possible to extrapolate a comprehensive system design and specification for the prospective in-vivo implementation of the technique.

3) Novelty of Research Methodology

The edge-illumination technique has been applied to the phase-contrast x-ray imaging of ex-vivo tissue samples [3][4], with its capabilities extended to tomographic imaging [5]. Building on these milestones, this project will exploit the high-resolution and texture of phase-contrast images to categorise tumour heterogeneity in lung cancer.

4) Alignment to EPSRC's strategies and research areas

This is a strongly interdisciplinary project, aligning with EPSRC themes across the domains of engineering, physical sciences, and healthcare technologies. Key areas of research are medical imaging, sensors and instrumentation, and artificial intelligence technologies. 5) Any companies or collaborators involved

The infrastructure available to this project includes a first commercial prototype X-ray phase-contrast scanner through the UCL-Nikon Prosperity Partnership, and two unique custom-built X-ray