Research Fellow in Deep Learning in Medical Image Computing & Modelling

Are you an early-, or mid-career researcher who wants to set the theoretical foundations that solve clinical and industrial problems?

Do you have a background in computer vision, medical image computing, machine and deep learning, biomedical engineering or computational multi-physics and multi-scale modelling?

Are you willing to take up the challenge to working across disciplines and on real-world data?

Are you passionate for combining computational algorithms, modelling and simulation in trailblazing research to create virtual patient populations and deliver in-silico trials in medical devices?

The Centre for Computational Imaging and Simulation Technologies in Biomedicine ( CISTIB ), within the Faculties of Engineering & Physical Sciences and Medicine & Health , involves various academics and their research groups. CISTIB focuses on algorithmic and applied research in the areas of computational imaging, machine learning, deep learning, and computational physiology modelling and simulation. CISTIB works in close cooperation with clinicians from various research centres from the University of Leeds and the academic hospitals of the Leeds Teaching Hospitals NHS Trust , one of the largest NHS Trusts in the UK. CISTIB is part of the Centres' conglomerate conforming the broader Leeds Centre for Responsive HealthTech Innovation in response to the recent Government Leeds City Region Science & Innovation Audit recognising the regional industrial R&D focus on MedTech.

CISTIB hosts a Royal Academy Chair in Emerging Technologies (2019-2029) to deliver INSILEX, a 10-year programme to undertake trailblazing research in Computational Medicine and In Silico Trials. INSILEX envisions a paradigm shift in medical device (MD) innovation where quantitative sciences are exploited to engineer MD designs, explicitly optimise clinical outcome carefully, and thoroughly test side-effects before being marketed. In-silico trials are essentially computer-based MD trials performed on populations of virtual patients. They use computer models/simulations to conceive, develop and assess devices with the intended clinical outcome explicitly optimised from the outset (a-priori) instead of being tested on humans (a-posteriori). This will include testing for potential risks to patients (side-effects) exhaustively exploring MD failure modes and operational uncertainties in-silico, before testing in live clinical trials. Advanced computer modelling will prove useful to predict how a device behaves when deployed across the general population or when used in new scenarios outreaching the primary prescriptions (device repurposing), helping to benefit the broadest possible target group without unintended consequences of side-effects and device interactions.

Within 10 years, we expect to have transformed MD design/evaluation by delivering these outcomes:

We are looking for Research Fellows supporting our work to address three main challenges we identified: 1) Build Virtual Patient Populations using probabilistic modelling; 2) Model device-tissue interactions through multi-physics, physiological modelling; 3) Develop efficient schemes to run ensembles of virtual experiments through accelerated numerical solvers and physics-informed machine learning. We identified cardiovascular medical devices as the first exemplar scenario (e.g. vascular stents, grafts and coils, valvular prostheses, etc.).

You can undertake innovative and high-impact research in one of the above areas: 1) deep learning for image analysis of cardiovascular population imaging (segmentation and modelling of cardiac chambers, valves, and vessels). These involve analysing datasets of several tens of thousands of images in an automatic manner; develop data harmonisation, image super-resolution, image imputation, generative image synthesis; and generative virtual population models. 2) modelling long-term response and failure models due to host organ-device interaction. Developing surrogate models for predicting long-term patient outcomes from technical device outcome measures. 3) Develop computational fluid dynamics, computational mechanics, and computational physiology methods accelerated using, amongst others, reduced-order models and physics-informed neural networks. Responsibilities will include developing new mathematical approaches to cardiovascular image analysis and computational physiology in the research areas outlined above; developing software implementation of these approaches in MULTI-X ( www.multi-x.org ), and communicating the research internally within the group, to external partners, and to the broader scientific community through journal publications and conference presentations. Using your knowledge of machine/deep learning, image analysis and computation, you will develop methods for highly automated and robust construction of image-based models of the cardiovascular system for subsequent multi-physics simulation of physiology. You will make technical and scientific contributions in line with the scientific goals of the underpinning projects from where the post draws its funding.

Research Fellows are academic researchers who can work independently, or as part of a research team, under the scientific leadership of a senior academic. Research Fellows have a strong academic profile including publication in top-rank peer-reviewed journals and conferences, excellent research enterprise and scholarship skills to conceive and pursue ground-breaking research questions, and they are able to manage graduate students and resources effectively setting direction, goals, a work plan, and monitoring progress. They are excellent problem solvers and are abreast of the advances in their field. They can serve as external liaison and maintain a network of local and external collaborators as required to deliver their goals.

You will hold a BSc/MSc (or equivalent) degree in relevant areas (i.e. physics, engineering, computer sciences, mathematics, or statistics), and a PhD (or thesis submitted and pending viva within 2 months) in a relevant computational domain. A track record in research commensurate with level of experience; in depth knowledge in statistics, physics, machine learning, and/or engineering and maths in the broad sense; excellent communication skills (oral and written); proficiency in computer programming (Python is a must as well as proficiency in at least MATLAB, C/C++ or other major programming language); extensive knowledge on deep learning environments (TensorFlow, Keras, and PyTorch) and GPU optimisation; ability to work independently, meet deadlines, take critique constructively, and excellent analytical skills are essential. Experience in medical imaging physics and analysis (e.g. MRI, CT, and US), and a good knowledge in statistics and/or Bayesian learning would be advantageous.

To explore the post further or for any queries you may have, please contact:

Professor Alex Frangi, School of Computing

Email: a.frangi@leeds.ac.uk