Research Topic: Developing AI/Machine Learning Methodology for Analysing Protein Organisation in Cells and Tissues as a Diagnostic Tool
Supervisors: Prof Michelle Peckham, Dr Alistair Curd, Prof Phil Quirke, Dr Joanna Leng, Dr Darren Tomlinson
About Umney: I studied an integrated masters degree in Physics at Manchester. In my final year project, I used machine learning to find target locations in 3D CT scans to determine facial asymmetry, with the goal being to generalise from adult to paediatric scans.
Project description: ∼40% of metastatic colorectal cancer (mCRC) patients receiving anti-EGFR (epidermal growth factor receptor) drugs panitumumab and cetuximab do not respond to treatment. Additional biomarkers such as expression levels of epiregulin (EREG) and amphiregulin (AREG) can help predict patient response to anti-EGFR agents in patients with wild-type RAS (rat sarcoma virus) oncogenes. EGFR/EREG organisation could be similarly incorporated to predict patient response to anti-EGFR therapies and could provide a better understanding of the EGFR stimulation pathway in mCRC. Single molecule localisation microscopy (SMLM) detects the positions of individual EGFR proteins and its ligands in tumour cells with very high resolution (∼20nm). My project will be looking at developing ML analysis pipelines of this SMLM data and examining how and whether it can be incorporated into routine medical care.