Prognostic modelling of coronary inflammation and plaque vulnerability using deep-learning of cardiac computed tomography imaging
LEAD SUPERVISOR: Professor Charalambos Antoniades, Radcliffe Department of Medicine
Co-supervisor: Professor Keith Channon, Radcliffe Department of Medicine
Commercial partner: Caristo Diagnostics, Oxford
Computed tomography coronary angiography (CTCA) is recommended as the first-line investigation in patients with suspected coronary artery disease. While current CTCA reporting focuses on qualitative evaluation of coronary stenosis, 60% of the heart attacks occur in people without significant coronary stenosis, since heart attack risk is driven by coronary inflammation rather than anatomical stenosis.
A novel technology that maps perivascular adipose tissue (PVAT) attenuation on CTCA, provides an accurate metric of coronary inflammation (perivascular Fat Attenuation Index -FAI), and it predicts heart attacks years in advance. Combining FAI with advanced plaque analytics to identify high-risk features plaques will provide a powerful approach to estimate cardiovascular risk.
This project will focus on the understanding of the role of coronary inflammation, using PVAT imaging biomarkers in the development of vulnerable atherosclerotic plaques that are prone to rupture, causing acute myocardial infarction. Specific objectives include:
a) Build a prognostic model that will calculate the patient’s absolute risk for fatal or non-fatal cardiac events, by using information from CTCA (coronary plaque and PVAT metrics), together with conventional risk factors and demographics (detecting the vulnerable patient). This will use the Oxford Risk Factors and Non Invasive imaging (ORFAN) study, and will rely on 65,000 CCTA scans collected from across the UK, linked with 15 years outcomes data from NHS Digital and National Institute for Cardiovascular Outcome Research (NICOR). This model will improve risk stratification and identify people in need for more aggressive medical management.
b) Generate new plaque-specific prognostic models for prediction of acute plaque rupture events (detecting the high-risk plaque). In the existing datasets of ORFAN, there are ~700 ST-elevated myocardial infarctions and ~1400 non ST-elevated myocardial infarctions, and the ruptured plaque is identified; so every event is attributed to a specific plaque, visualized on CTCA several years before it ruptured. This will allow detection of high-risk plaques, prone to rupture, enabling local intervention (e.g. PCI) in a timely manner, with a potential to prevent heart attacks.
A key strategic priority of the MRC is prevention and early detection to improve population health. The emphasis of this project is to harness the interpretation of CTCA which is already in widespread use as a first-line test for diagnosing coronary artery disease to enhance risk profiling for future cardiovascular events above and beyond the current models that are based on risk factors (e.g. age, gender, hypertension, hyperlipidaemia, diabetes and smoking status). Caristo diagnostics is a pioneer SME in developing the technology that quantify coronary artery inflammation using PVAT biomarker on CTCA. Integrating with advanced atherosclerotic plaque analysis will deliver a comprehensive tool to identify vulnerable individuals, allowing timely and tailored treatment plan to modify the risk of cardiovascular events.
Apply using course: DPhil in Medical Sciences
January 2023 update:
Applications for this iCASE project (for October 2023 entry) are no longer accepted.
