Skip to the content.

Development of digital twins using physics-informed deep generative learning

Project ID: 2531ad1522

(You will need this ID for your application)

Research Theme: Healthcare Technologies

UCL Lead department: Division of Medicine

Department Website

Lead Supervisor: Simon Walker-Samuel

Project Summary:

The drive towards personalised medicine requires an understanding of the individual characteristics of each patient and their individual disease. Using structural and functional information from imaging technologies such as MRI, x-ray CT and optical coherence tomography (OCT), we have been developing spatial and temporal biophysical simulations of a range of phenomena, including tumour drug delivery and retinal disease progression. Combining this with recent advances in deep generative learning (a form of AI) has enabled us to couple physics-based simulations to real-world clinical data on an individual basis (for example, Brown et al, Nature Communications 15: 6859 (2024)). This project will aim to further develop this approach by: 1) enriching our physics-based models using structural data from high-resolution three-dimensional microscopy; 2) developing methods for linking these models to clinical data using deep generative learning (e.g. image style transfer with latent diffusion models); 3) investigate and validate this approach using established clinical image libraries (e.g. retinal OCT). The outcomes of a successful project will include new approaches to create personalised digital twins (building on our previous work), whose accuracy and reliability have been rigorously assessed, with a view towards translation into the clinic. Successful candidates would have a background in a numerical discipline (e.g. physics, mathematics, computer science) and with experience of coding (ideally python, c++ or Matlab), although applications will be considered on a case-by-case basis. Students will be provided with extensive training in each of the supporting technologies, and will be based in the Centre for Computational Medicine, which aims to provide a diverse, creative and supportive environment that focusses on the development of all students.