Computer Science and Mathematics

Computational Imaging is a major area of focus with groups developing new algorithms for image analysis, multi‐modal image fusion, motion correction, quantification and modelling of molecular imaging tracers, and descriptive atlases of anatomy and physiology. We are developing virtual histology, using “dictionaries” of tissue annotations mapped to MR images. Computational imaging techniques are being applied to cancer, neuro‐development/ degeneration as well as cardiovascular diseases, and used to extract clinically meaningful information for improved diagnostics, patient stratification and outcome prediction.

Deep Learning (DL) is an emerging field within machine learning that using Big Data to design artificial intelligence algorithms. We are designing programmes to use hundreds of thousands of historical records from partner NHS trusts. We are applying DL in a number of ways; using DL algorithms to automatically detect interesting regions of images to study disease; applying DL to link genetic/epigenetic data to imaging phenotypes; applying DL to automatically design the best experiment to create a specific brain state in a given subject; and applying DL to characterise tumour heterogeneity from historical data in lung and prostate cancer.

Computational Modelling: we are developing techniques to simulate the function of organs and systems, from multi‐modality diagnostic data. These models allow personalisation to a patient’s anatomy and physiology and also allows us to predict the patient’s response to particular treatments. We can generate cohorts of virtual patients for prospective in‐silico clinical trials.