Computational Medicine

We use biomedical imaging to guide computational analysis, alongside developing novel, cutting-edge imaging techniques with a focus on translation and providing patient benefit. Ongoing project areas include:

Magnetic Resonance Imaging (MRI) of Cancer

We are developing quantitative MRI techniques to characterise the tumour microenvironment and predict drug delivery and treatment response. This includes techniques such as diffusion MRI modelling of tumour microstructure to non-invasively measure cell size and arterial spin labelling to measure vascular perfusion. (Recent paper)

Imaging Transplanted Organs as Model Systems

We are passionate about understanding, validating and translating new imaging approaches, and are developing methods for trialling new technologies in transplanted human organs that are maintained by mechanical perfusion. This provides a novel platform to investigate real tumours in situ and how they interact with imaging biophysics (collaboration with Royal Free Hospital and WEISS).

Optical Imaging

We develop new optical imaging solutions such as high-resolution episcopic microscopy (HREM) and light-sheet microscopy. This allows us to create large-scale digital tissue models (‘digital twins’) that can be used for mathematical modelling and for better understanding biomedical imaging techniques.

HiP-CT

Mapping human organs at multiple scales using synchrotron radiation. (Recent paper)

We use mathematical modelling to explore and quantify the relationship between function and form.

Blood Flow and Solid Mechanics in Tumours

REANIMATE is our framework for combining three-dimensional microscopy of tumours and in vivo imaging, to model blood flow, and interstitial fluid dynamics to better understand and optimise drug delivery to tumours. The interaction between fluid mechanics and solid mechanics is an active area of investigation, particularly in relation to metastatic growth in the spine (collaboration with Leeds University). (Recent paper)

Blood Vessel Network Simulation

We have developed several platforms for modelling normal blood vessel networks, using biophysical principles such as Murray’s Law, alongside models of aberrant growth such as angiogenesis in tumours.

OncoEng

An EPSRC-funded project in collaboration with the University of Leeds, aiming to predict fracture risk in spinal metastases. (Project website: OncoEng).

Retinal modelling

We are creating detailed, large-scale models of the human retina, using ophthalmology image data, to better understand the link between vascular delivery and metabolic demand (collaboration with Moorfields Eye Hospital and Institute of Ophthalmology). Understanding the cause of loss of vascular perfusion in diabetic retinopathy is a specific motivation and its impact on vision. 

Liver Perfusion

Using a range of MRI approaches, we are developing large-scale models of liver vascular perfusion, particularly during the growth of tumours, and investigating the delivery of therapeutic nanoparticles (collaboration with Royal Free Hospital and UCL Chemistry).

Physics informed deep learning

Our imaging and modelling work combines physics-informed deep learning, in which mathematical models are used to train generative learning algorithms. For example, working with colleagues at Moorfields Eye Hospital, we are developing deep generative modelling approaches to segment and classify images from the retina.

Supervised learning

We have developed a large imaging database of vascular imaging data with manual labels, which forms the basis of our tUbe-Net framework for segmenting blood vessel networks with minimal manual labelling (more on GitHub).