Studentships in Magnetic Resonance & Precision Imaging

Supervisors

Dorothee Auer (Clinical Professor of Neuroimaging)

Richard Bowtell (Professor of MR Physics)

Adam Berrington (Assistant Professor in MR Spectroscopy)

Project proposal

Brain energy metabolism in physiological conditions is uniquely dependent on glucose utilization with ATP generation through glycolysis and the TCA cycle. Metabolic reprogramming is the physio-pathological response to metabolic stress which is well studied in cancer but becomes increasingly recognised in a range of brain diseases such as inflammation (‘immunometabolism of microglia’), inborn and acquired metabolic encephalopathies, traumatic and ischaemic injuries and importantly neurodegenerative disorders. A better understanding and non-invasive detection of the pathogenetic and compensatory metabolic mechanisms is urgently needed to identify novel treatment targets and monitor targeted interventions. The only clinical metabolic imaging tool FDG-PET cannot address this need due to limited availability, high cost, radiation exposure and foremost inability to study downstream metabolic products. Novel advanced MR methods exploiting ultrahigh-field capabilities and other nuclei combined with safe labels (e.g. deuterated glucose) offer huge potential to overcome these limitations for in depth characterisation of neurometabolic underpinning of brain disorders. We developed a non-invasive protocol including an oral (D7 and D2-deuterated) glucose drink followed by serial deuterium MR imaging at 7T that allows to dynamically map glucose uptake, glycolytic activity (lactate production) and TCA activity (glutamate/glutamine increase).

This PhD project aims to exploit this novel technological platform for characterisation of metabolic abnormalities in brain disorders focusing on the detection of metabolic reprogramming in non-neoplastic conditions (e.g. dementia, neuroinflammation). Clinical proof-of-concept studies will focus on conditions with candidate metabolic targets for future clinical trials. For easier clinical translation, proton-based detection methods will be developed. The project is particularly well suited for candidates with a MR Physics, Biochemistry or Neuroscience background and strong interest in clinical translation.

Supervisors 

Richard Bowtell (Professor of MR Physics)

Akram Hosseini (Honorary Associate Professor of Physics)

Simon Paine (Associate Professor of Medicine)

Penny Gowland (Professor of Physics)

Olivier Mougin (Senior Research Fellow in Physics)

Project proposal  

The aim of this project is to better characterise hippocampal changes in Alzheimer’s disease (AD).

The student will develop and optimize quantitative MRI metrics of key contrasts in the hippocampus for AD in vivo. Although there has been previous work in this area, particularly looking at iron, the link between iron and amyloid deposition remains to be fully described, and we aim to provide markers for both.

However, we will need to validate our markers, and therefore, the student will compare in vivo scans to whole brain post mortem MRI scans which will in turn be compared directly to histopathology (performed by others). This will allow us to correlate iron deposition (measured using quantitative susceptibility mapping (QSM) in MRI) and the presence of amyloid (measured through potential new MRI markers), with histological measurements. This work will be critical for establishing biomarkers of neurodegeneration.

It is also anticipated that the student will undertake some of the first studies at 11.7T to investigate the use of MRI for detecting the distribution of cell types, myelin and iron, and to better understand the origins of the contrasts used in brain MRI at high field, with a goal of producing the first comparison of histopathology to 11.7T MRI.

The student will gain detailed skills and knowledge in MRI physics, image analysis, basic neuroscience and excellent experimental technique

Supervisors

Penny Gowland (Professor of Physics)

Steffi Thust (Clinical Associate Professor in Medicine and Health Sciences)

Susan Francis (Professor of Physics)

Nicholas Blockley (Assistant Professor)

Project proposal

Brain tumours represent a major cause of cancer deaths, including in young persons. Magnetic resonance imaging (MRI) is fundamental to characterise neoplasms and plan treatment, which differs by tumour type. Complex challenges of genetic tumour classification, visualisation of disease spread and recognizing viable brain cancer cells after treatment are not yet solved by clinical MRI.

Technical advances at Sir Peter Mansfield Imaging Centre (SPMIC) have created new opportunities to probe brain tissue microstructure. Ultrahigh magnetic field (UHF) MRI offers substantially greater spatial resolution with potential advantages for subtle detail depiction and physiometabolic image signal amplification.

The aim of this research is to develop an innovative UHF MRI (7T) protocol for precision mapping of brain tumours. This will include a selection of high-resolution anatomical, physiological (perfusion, vessel size imaging, oxygenation, susceptibility) and metabolic (chemical exchange, oncometabolite spectroscopy, sodium imaging) techniques for multimodal analysis.

The PhD will focus on novel sequence application, technical development, and post-processing. Where valuable, artificial intelligence (AI) methodology will be integrated into image acquisitions and computational analyses. One goal of this work is to initiate a pathway for clinical translation onto a 3T MRI platform at Nottingham University Hospitals. The process of methods selection and translation will be enhanced through correlative tissue molecular studies in collaboration with the Biodiscovery Institute (BDI) and Nottingham Nanoscience & Nanotechnology Centre (NMRC), University of Nottingham. 

Supervisors

Andrew French (Professor of Computer Science)

Nikola Sprigg (Professor of Stroke Medicine)

Stefan Pszczolkowski (Researcher, Medical Image Analysis)

Rob Dineen (Professor of Neuroradiology)

Project proposal

Medical imaging, more so than other fields of image analysis, suffers from two critical limitations when developing AI solutions for image interpretation:

1. Limitations of training data, particularly given the expertise required for labelling, and the tight time constraints on those clinical experts required to carry out that labelling. This is particularly true for segmentation tasks, where labels require careful detailing of image regions, rather than course, image-level categorisation.
2. Variations between acquisition devices. Even within the same modality, it is recognised that devices from different manufacturers, or even the same device calibrated differently or used as part of a different protocol, can produce images which appear different to machine learning systems.

These are well-recognised as critical challenges to be addressed in order for AI approaches to become more rapidly adopted in medical image analysis. Our novel and timely approach to solving these problems will use Generative AI-based approaches based on real-world medical images from large multicenter clinical trials to build new ways of more efficiently training machine learning systems with minimal amounts of expert-created training data, and which permit systems trained on images from one device to adapt to the nuances of images captures from a different devices.

Supervisors

Xin Chen (Associate Professor in Computer Science)

Steffi Thust (Clinical Associate Professor in Medicine and Health Sciences)

Project proposal

This research aims to develop artificial intelligence (AI) methods for lesion detection, segmentation and diagnosis. Key aims are to maximise identification of early brain cancer (glioblastoma) stages (where AI literature outputs report on late disease features) and to distinguish a wider range of conditions including mimics. Ethical approval has been obtained for a study of 5000 pseudonymised patient datasets from Nottingham University Hospitals.

The PhD will focus on 4 areas of research: 1. Brain abnormality detection, 2. Automated segmentation with optional function for human interaction (review and manual correction), 3. Classification and 4. Integration of new parameters (diffusion and age) into predictive modelling, which is lacking from published AI developments. The PhD candidate will develop fundamental AI algorithms to address challenging issues, such as segmentation of small objects, generalisation of machine learning models and integration of AI into clinical practice. This project will link precision radiomics with specific diagnostic needs and a clear vision to clinical translation.

Supervisors

Paul Morgan (Chair in Medical Physics)

Michael Chappell (Professor of Biomedical Imaging)

Steffi Thust (Clinical Associate Professor in Medicine and Health Sciences)

Akram Hosseini (Associate Professor of Physics)

Project proposal

Measurement of blood perfusion in the brain is used clinically for a number of conditions, especially brain tumours. This typically involves an injection of contrast agent during the perfusion scan. Research MRI studies also measure brain perfusion, often using the Arterial Spin Labelling (ASL) technique without contrast agent, although this is less common in clinical use. Studies are approved to use these techniques in brain tumours and dementia, but there is a lack of standardisation across contrast and ASL techniques, or clear definition of which technique best shows which diagnostic parameters. In addition, perfusion MR images are usually relatively low resolution, whereas clinically perfusion values are desirable at higher spatial resolutions; the uncertainty in perfusion measurements in pathology using ASL at different resolutions will also be assessed.

This project seeks to clarify these parameters, by acquiring various perfusion techniques on the same patients, including ASL to various spatial resolutions and analyse them using the same software, to reach an overarching consensus on the appropriate use of these perfusion techniques for neuro perfusion MRI.

Supervisors 

Stephen Coombes (Professor of Applied Mathematics)

Stam Sotiropoulos (Professor of Computational Neuroimaging)

Marcus Kasier (Professor of Neuroinformatics)

Project proposal  

Myelin pathology has long been associated with diseases such as multiple sclerosis (demyelination), and more recently with psychiatric and neurodegenerative disorders including depression and schizophrenia (where structural differences in white matter networks are manifest). It has only relatively recently been established that myelin is also modifiable by experience and can affect information processing by regulating the velocity of signal transmission to produce synchronous arrival of synaptic inputs between distant (and multiple) cortical regions. Indeed, myelin plasticity is increasingly being seen as a complementary partner to synaptic plasticity and, as well as being important to nervous system development, it has a major role to play in complex information processing tasks that involve coupling and synchrony among different brain regions. Importantly, myelin plasticity is a putative mechanism for non-invasive neuromodulation. This project will build a new computational framework for biologically motivated neural networks to help understand the important contribution that activity-dependent regulation of myelination can make to patterns of rhythmic activity known to subserve important aspects of large-scale brain dynamics and its dysfunction. It will:

1. Combine perspectives from neural mass and network modelling and develop a new set of computational tools able to unravel the contributions of space-dependent axonal delays to large-scale spatio-temporal patterning of brain activity; 
2. Develop new computational models for myelin based plasticity and analyse their consequences for network timing.