Centres for Doctoral Training

Modelling Morphology Evolution in Hepatocellular Carcinoma with Generative AI

Supervisors:

Ram DasGupta, MVLS/School of Cancer Sciences

Ke Yuan, MVLS/School Cancer Sciences

Paul Henderson, CoSE, School of Computing Science

 

PhD Project Summary:  

Liver cancer is one of the deadliest cancers, with an abysmal 5 year median survival of <15%. This is largely due to its ability to rapidly evolve under therapeutic selection into refractory, metastatic disease. Our research team has recently developed an AI methodology that can automatically detect key morphological patterns from tumour pathology slides. What remains unknown is how these patterns, and the communities of cells within them, change when tumours are exposed to therapy.

This PhD project aims to investigate how liver cancer morphology evolves under treatment, over time. Working with a unique collection of patient samples taken both before and after therapy, you will apply our state-of-the-art AI methodologies to chart how tumour structures evolve over time. You will then integrate these findings with spatial transcriptomics, a powerful technology that measures which genes are actively expressed in different regions of the tumour while preserving their spatial context.

By combining AI with spatial biology, you will uncover the molecular programs that drive therapy resistance and develop the first computational model to simulate this evolutionary process. This interdisciplinary project brings together cancer biology, pathology, and AI with the ultimate goal of identifying new ways to predict treatment failure and design more effective therapies.