2023-24-project-catalogue

###Tera-scale deep learning for knowledge extraction from Hierarchical Phase-Contrast Tomography

Project ID: 2228bd1046 (You will need this ID for your application)

Research Theme: Healthcare Technologies

UCL Lead department: Computer Science

Department Website

Lead Supervisor: Danny Alexander

Project Summary:

This project develops new deep learning techniques for exploitation of a recent imaging innovation call Hierarchical Phase Contrast Tomography (HiP-CT): https://mecheng.ucl.ac.uk/HiP-CT. HiP-CT offers unique insight into the microstructural components of tissue, e.g. airways, blood vessels, cells, etc. by achieving unprecedented resolution and field of view. However, the vast amount of data it produces requires new image analysis tools for identification, segmentation, and classification of tissue components. The project puts those tools in place and uses them to provide new understanding of the contrasts we observe in clinical imaging modalities such as MRI and CT. This will enhance diagnostic power and disease understanding in key contemporary medical challenges such as prostate cancer and Alzheimer’s diseases. HiP-CT enables human organs to be scanned hierarchically from the whole organ at 20-8μm/voxel down to 1μm/voxel in local regions anywhere within the intact organ. It relies on the exceptional coherence and high energy provided by the European Synchrotron Research Facility’s Extremely Bright Source (ESRF-EBS).

The large data sets (up to 1TB) that this produces require advanced tools for meaningful interpretation and analysis. Deep learning (DL) has revolutionised medical imaging providing quantum leaps in performance of standard image analysis tasks such as segmentation, classification or super-resolution. It provides the key to extracting the new knowledge from this emerging imaging modality and thus enabling major advances in some of the biggest challenges facing 21st Century medicine. Applicants should have a degree in Physics, Mathematics or Computer Science (or a related subject), awarded at 2:1 level (UK system or equivalent) or above. Knowledge of basic image processing and computer programming skills are required.