Causal modelling of disease progression in medical images and associated clinical data
4-year PhD studentship: ** Now Closed**
9 March 2021
4-year PhD studentship: Causal modelling of disease progression in medical images and associated clinical data
Prof Daniel C. Alexander, UCL Centre for Medical Image Computing
Dr Daniel Coelho de Castro, Microsoft Research Cambridge
A Microsoft PhD Scholarship is available for a 4-year PhD studentship to be based in the UCL Centre for Medical Image Computing (CMIC). The studentships comes with a generous stipend approximately at the Wellcome Trust rates (the maximum possible for PhD studentships).
The successful candidate will align with the UCL CDT in Intelligent, Integrated Imaging in Healthcare (i4health) cohort and benefit from the activities and events organised by the centre. They will have a co-supervisor from Microsoft Research Cambridge, and will have the opportunity for a short-term placement within the medical image analysis team at MSRC.
The motivation for the work is to better understand Alzheimer’s disease, and ultimately other long-term chronic conditions, through computational modelling and machine learning informed by large imaging data sets. That understanding informs the development, demonstration, and eventual deployment of new disease-modifying treatments that are currently unavailable for these devastating conditions. Specifically, the project combines disease progression modelling techniques developed by the POND group at UCL, such as the event-based model (Fonteijn, Neuroimage 2012) and SuStaIn (Young, Nature Communications 2018), with deep structural causal models developed by researchers at Microsoft (see for example Pawlowski, NeurIPS 2020). This enables newly expressive spatio-temporal models of change during long-term chronic illness, uniquely allowing us to visualise the effects of interventions (how will this patient’s prognosis change if they stop smoking?) and counterfactual alterations (how would this patient look now if they had stopped smoking ten years ago?) on the disease trajectory. The project focusses on several underlying technical challenges: how to exploit structural causal models to constrain inference of long-term disease trajectory from irregularly sampled, short-term longitudinal data sets; how to synthesize image and associated clinical data from the model realistically enough to support clinical inference. We focus initially on Alzheimer’s disease data sets, as the necessary access to data sets and clinical expertise are readily available at UCL through projects such as EuroPOND and E-DADS
- Build causal statistical models that are able not only of forecasting disease progression in the clinical covariates according to existing hypotheses, but also of synthesising realistic images that plausibly match the conditioning attributes.
- Establish effective techniques to optimise the parameters of such temporal models from irregularly sampled, unbalanced, multimodal longitudinal data, or even from plain cross-sectional datasets.
- Uncover the effects of various clinical factors (demographics, lifestyle, genetics, etc.) on patient trajectories for applications such as neurological or respiratory diseases
Person specification & requirements:
The ideal applicant will have a strong mathematical and/or computational background and a keen interest in effecting meaningful change to patients living with chronic diseases such as dementia.
Candidates should have experience and interest in developing machine learning methods (statistical models) for medical image and data analysis. Experience with any of image analysis, big clinical data analysis, Bayesian modelling, longitudinal statistical modelling, and causal inference would be advantageous but not essential.
Candidates must have a UK (or international equivalent) first class or 2:1 honours degree and an MSc in physics, computer science, mathematics, engineering, or a comparable subject.
This is an EPSRC CASE award. For funding eligibility please consult the following webpage: Guidance on EPSRC student eligibility
To apply please send your CV and expression of interest to: firstname.lastname@example.org
Closing Date: 15th April 2021
Interviews: 10th May 2021