Developing an AI-based precision classifier of radiotherapy resistance in gastrointestinal cancers

Project ID: 2228cd1327 (You will need this ID for your application)

UCL Lead department: Division of Biosciences

Project Summary:

• Why this research is important:

Radiotherapy is one of the 4 pillars of curative cancer treatments, but often results in the development of resistance, primarily orchestrated by elusive “persister” cells that are challenging to detect. Crafting a precision classifier for these treatment-resistant persister cells would allow us to identify patients where radiation therapy can be personalised.

• Who you will be working with: Dr Maria Secrier (GEE, Computational Cancer Biology), Prof Maria Hawkins (Medical Physics, Radiation Oncology)

• What you will be doing:

Your PhD will make use of 1500 annotated tumour tissue histopathology slides alongside matched RNA sequencing data from patients profiled before and after radiation therapy and associated clinical outcomes, sourced from CRUK-funded ARISTOTLE (rectal cancer) and PROTIEUS (oesophageal) clinical trials and specimens from standard of care subjects treated in the UCLH. You will leverage artificial intelligence and graph theory on this data trove to probe persister cells and their resistance dynamics within tumours after radiation. Employing a bespoke RNA-based signature for persister cells and the Multiple-Instance Learning framework HistoMIL developed by the Secrier lab, you will develop a robust deep learning classifier aiming to gauge therapeutic resistance in digital pathology slides. You will further analyse the immune cell content of these tissues utilising graph theory and graph neural networks to understand tumour-immune cell dependencies that could be exploited therapeutically. This exploration of persister cells and their environment in the context of radiotherapy will provide insights into therapy efficacy and unveil novel resistance biomarkers, alongside potential radiation-drug synergies.

Skills gained: analysing and interpreting digital pathology data using AI and graph theory approaches, in-depth understanding of cancer biology

• Who we are looking for:

An enthusiastic individual with a background in mathematics, statistics, computer science, engineering or a similar discipline, and a keen interest in cancer research.