Overview
Age-related macular degeneration (AMD) is the leading cause of blindness globally. Current treatments involve the injection of anti-vascular endothelial growth factor. However, many individuals do not respond to treatment, with regular visits required placing a large burden on healthcare systems and patients. The longitudinal follow-up scans, however, provide a rich dataset to develop prediction tools to determine responders and non-responders and to tailor treatments.
About this opportunity
This project will use longitudinal patient data at the primary institute to develop digital twins of the retinal microcirculation for individual patients. These computational twins will be used to help predict outcomes of treatment and cardiovascular adverse events.
Objectives
WP1: Curating an AMD patient dataset
As part of an ongoing study at the primary institution, the PhD candidate will attend the University hospital to extract AMD patient data and link it to their retinal scans. There are 400 eyes currently recruited into the study, hence this first work package will expand this number and connect the patient scans to the electronic health records.
WP2: Developing a Digital Twin pipeline
Building on previous work in generating retinal vascular computational simulations, the candidate will develop a pipeline that uses patient scans, segments the vasculature, and recreates that patients’ retinal morphology on a computer (a digital twin). Blood flow and oxygen transport simulations will be conducted to extract in silico biomarkers that are not readily available in vivo. Longitudinal patient scans will be used to update the patient digital twin.
WP3: Predictions with the Digital Twin
The digital twin, along with patient data from the electronic health record, will be used to predict treatment to response and any cardiovascular adverse events for a patient. This will be done with a multi-modal machine learning model incorporating patient data, patient scans, and the in silico biomarkers from the digital twin.
Novelty
The novelty of this project lies in the use of in silico modelling and longitudinal clinical data to build digital twins to provide personalised dynamic risk prediction for individuals.
Timeliness
The eye is a window to vascular health. With an increasing multi-morbid and ageing population, developing digital twins to predict outcomes will improve risk prediction and health outcomes.
Experimental Approach
This PhD incorporates image and patient data analysis, deep learning, mathematical modelling, and software development.
The work to be undertaken will be conducted at the Department of Cardiovascular and Metabolic Medicine, the Department of Eye and Vision Sciences, and the Liverpool Centre for Cardiovascular Science as a collaboration between biomedical engineers (Dr El-Bouri), deep learning specialists (Prof. Zheng) and clinical experts in diabetes and ophthalmology (Dr Alam, Dr Madhusudhan).
Further reading
Hernandez, Rémi J., Savita Madhusudhan, Yalin Zheng, and Wahbi K. El-Bouri. “Linking Vascular Structure and Function: Image-Based Virtual Populations of the Retina.” Investigative Ophthalmology & Visual Science 65, no. 4 (April 1, 2024): 40–40. https://doi.org/10.1167/IOVS.65.4.40.
Hernandez, Rémi J, and Wahbi K El-Bouri. “Microvascular Retinal Digital Twins from Non-Invasive Clinical Images” edited by Lei Li, Viktor Jirsa, Jianfeng Feng, Jun Deng, Luca Dede’, Sora An, Yilin Lyu, and Xiaoyue Liu, 12–22. Cham: Springer Nature Switzerland, 2026.
Hernandez, Rémi J, Paul A Roberts, and Wahbi K El-Bouri. “Advancing Treatment of Retinal Disease through in Silico Trials.” Progress in Biomedical Engineering 5, no. 2 (2023): 22002. https://doi.org/10.1088/2516-1091/acc8a9.
Hernandez, Rémi J., Wahbi K. El-Bouri, Savita Madhusudhan, and Yalin Zheng. “AI and the Eye – Integrating Deep Learning and in Silico Simulations to Optimise Diagnosis and Treatment of Wet Macular Degeneration.” MedRxiv, February 14, 2024, 2024.02.13.23299445. https://doi.org/10.1101/2024.02.13.23299445.
Brown, Emmeline E., Andrew A. Guy, Natalie A. Holroyd, Paul W. Sweeney, Lucie Gourmet, Hannah Coleman, Claire Walsh, et al. “Physics-Informed Deep Generative Learning for Quantitative Assessment of the Retina.” Nature Communications 2024 15:1 15, no. 1 (August 10, 2024): 1–14. https://doi.org/10.1038/s41467-024-50911-y.