AI-enhanced cardiovascular imaging using machine learning at population scale

external supervisors

Qiang Zhang, Associate Professor of AI in Cardiovascular Imaging Radcliffe Department of Medicine

Vanessa Ferreira, Professor of Cardiovascular Medicine, Radcliffe Department of Medicine

BACKGROUND

This DPhil project focuses on leveraging generative AI models to enhance cardiac MR images in the UK Biobank, a comprehensive biomedical database.

For heart disease, the most informative imaging is that of myocardial fibrosis (scarring) using late gadolinium enhancement of cardiac MRI. However, this is not available in UK Biobank, due to the need for intravenous contrast injections in performing fibrosis imaging.

In our pilot study we have demonstrated that generative AI models can explore diagnostic information in contrast-free MRI and produce AI-enhanced imaging matching the current clinical standard for myocardial fibrosis assessment. AI is effectively employed as a ‘virtual contrast’. We aim to further develop and deploy this concept for UK Biobank to significantly augment the biobank cardiac imaging database.

This DPhil project aims to:

• develop and deploy AI enhancement for UK Biobank cardiac MR materials
• study the association of AI-enhanced cardiac imaging in various clinical conditions with adverse outcomes in UK Biobank, and identify the most predictive imaging features
• study the association of myocardial injuries identified by AI-enhanced imaging with conditions of multiple organs such as brain and liver using advanced machine learning models.

RESEARCH EXPERIENCE, RESEARCH METHODS AND TRAINING

You will benefit from an interdisciplinary core supervision team of machine learning scientist, biomedical statistician and cardiologist, and develop and validate deep generative AI models at population scale. Skills training includes machine learning (deep learning), cardiac imaging and data analysis for large population studies.

The project is in partnership with Oxford Centre for Clinical Magnetic Resonance Research (OCMR) which supports expertise in cardiac MRI and cardiology. 

Research works are expected to lead to several novel publications at the intersection of AI and healthcare.

FIELD WORK, SECONDMENTS, INDUSTRY PLACEMENTS AND TRAINING

You will be primarily based at Big Data Institute. Additionally, you will have access to facilities, mentorship and networking at RDM Division of Cardiovascular Medicine. You will have the chance to observe and learn about real-world clinical MR scans at OCMR. You will contribute to the collaborations between the two departments and the integration of large-scale cardiovascular image analysis with machine learning methods.

PROSPECTIVE  STUDENT

The ideal candidate will have a degree in computer science and/or machine learning, and have research experience in medical imaging.