Predicting the future of heart health: data-economic deep learning for the prediction of acute coronary syndromes from cardiac CT images
Interview date: Tuesday 27th May 2025
Background
Coronary heart disease (CHD) remains a leading cause of mortality worldwide [1,2] with the UK county of Lincolnshire experiencing a particularly high prevalence relative to the rest of the country [3]. Early detection and prevention strategies are therefore crucial in combating this widespread health issue. Computed tomography coronary angiography (CTCA) has become the de facto gold standard for evaluating CHD, providing detailed images of the coronary arteries and the ability to quantify both narrowing (stenosis) and atherosclerotic plaques. Notably, non-calcified plaques are strong indicators of future acute coronary syndromes (ACS), such as heart attacks, making their accurate characterisation vital for risk assessment [4].
Artificial intelligence (AI) has shown immense promise in analysing medical images to predict long-term health outcomes. However, the application of AI, particularly deep learning, to the detailed analysis of coronary plaque characteristics for predicting ACS remains relatively underexplored. A significant hurdle lies in the substantial computational resources and large datasets typically required to train deep neural networks effectively. This presents a challenge in medical imaging, where data acquisition, annotation, and standardisation across different healthcare settings can be complex and resource-intensive. Furthermore, ensuring that AI models can generalise well across diverse patient populations and different imaging protocols is critical for their real-world clinical utility.
The project
This innovative, EPSRC-funded PhD project aims to overcome these limitations by developing novel AI approaches for ACS risk prediction with a specific focus on data-economic solutions. The research will investigate and leverage advanced techniques such as domain generalisation, knowledge distillation, and self-supervised learning to create models that can achieve high predictive accuracy while requiring significantly fewer labelled training samples. This approach addresses the practical challenges of data scarcity and paves the way for more cost-effective and widely applicable AI solutions in cardiovascular care.
Working in close collaboration with clinical stakeholders at Lincolnshire Heart Centre (United Lincolnshire Teaching Hospitals Trust), the successful candidate will have access to valuable clinical data and expertise. The project will involve developing AI algorithms capable of learning from limited labelled data by transferring knowledge from related imaging domains and by effectively incorporating structured clinical variables, such as patient demographics, existing health conditions, and known risk factors, to enhance the predictive power, generalisability and fairness of the models.
Candidate requirements
We are seeking highly motivated candidates with a strong (minimum 2:1 Bachelor’s degree, or international equivalent) background in computer science, machine learning, biomedical engineering, or a related quantitative discipline. Experience with medical image analysis is desirable but not essential. If you are passionate about applying advanced AI techniques to solve critical healthcare challenges and are eager to contribute to research with direct translational potential, we encourage you to apply.
Research environment
Dr James Brown is an Associate Professor in Computer Science and the Deputy Director of Research in the School of Engineering & Physical Sciences. Dr Wenting Duan is a Senior Lecturer in Computer Science, and the programme leader for MSc Computer Science. Drs Brown & Duan are the Director and Deputy Director of the Laboratory of Vision Engineering (LoVE), respectively (https://www.visioneng.org.uk). Our vibrant, interdisciplinary team currently comprises seven academics, seven PhD students, and three postdoctoral research associates, specialising in the analysis and understanding of image and video data with applications spanning the medical, life and environmental sciences.
References
[1] https://www.nhs.uk/conditions/coronary-heart-disease
[2] https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death
[3] https://lhih.org.uk/jsna/live-well/cardiovascular-disease
[4] Nerlekar, N., Ha, F.J., Cheshire, C., Rashid, H., Cameron, J.D., Wong, D.T., Seneviratne, S. and Brown, A.J., 2018. Computed tomographic coronary angiography–derived plaque characteristics predict major adverse cardiovascular events: a systematic review and meta-analysis. Circulation: Cardiovascular Imaging, 11(1), p.e006973.
To apply for this project, visit our Apply Now page.