Novel histomorphological predictive biomarkers of prognosis and therapy response in pancreatic and lung cancers.

Supervisors:

Dr Ke Yuan, School of Cancer Sciences & School of Computing Science
Prof David Chang, School of Cancer Sciences
Prof John Le Quesne, School of Cancer Sciences
Mr Christopher Walsh, School of Cancer Sciences

Summary:

This project aims to identify novel histomorphological biomarkers for predicting prognosis and therapeutic response in pancreatic and lung cancers. Using cutting-edge artificial intelligence and machine learning techniques, we will integrate the information in haematoxylin and eosin (H&E) and multiplex immunofluorescence (mIF) whole slide images to enable unparalleled understanding of the tumour and cell histology and tissue microenvironment.

This project combines a tile-based approach, which groups histopathological tiles from whole slide images into clusters based on morphology, and a single-cell approach, which uses self-supervised learning to understand nuclear features and cell microenvironments. Together, these methods will allow us to uncover histological patterns and cellular morphologies that impact disease progression and treatment response.

We will leverage large, well-curated datasets of pancreatic and lung cancer patients with accompanying mIF, genomic and transcriptomic data. The integration of H&E and mIF images will enable the discovery of clinically meaningful biomarkers that can be easily applied to routine clinical pathology, enhancing their translatability. This exciting research could lead to personalised treatment strategies and improved prognostic models, directly impacting patient care.