Research
Overarching research question
Can we use routine H&E images to develop low-cost AI-defined molecular tests and use them for tailoring cancer treatments in a global setting?
PARP inhibitor therapy has significantly improved patients survival in ovarian cancer. Our initial focus is on a key biomarker guiding treatment decisions of PARP inhibitor therapy known as homologous recombination deficiency (HRD).
AAIMC-OC Pilot objectives
- To refine AI algorithms to evaluate HRD and other molecular characteristics in ovarian cancer from H&E whole-slide images (WSIs) using a multinational dataset
- To generate a matched cohort of images using different imaging systems and at different magnifications relevant for different resources settings to evaluate model flexibility of the established AI algorithms
- To evaluate the technical feasibility to deploy already-trained AI-defined molecular tests in poor-resource settings using a privacy-preserving federated learning platform

Our work so far
Now:
We have established international collaboration agreements and licenses to conduct research in low-and-middle-income-country settings, multinational datasets and early federated learning infrastructure test.
Please also see our Open Science Framework (OSF) Project Page for our latest output and activities.
What’s next?
- Development of interoperable data systems and multi site evaluation of AI diagnostics
- Currently our funding does not support gold-standard comparator molecular tests in low-and-middle-income countries. This will be important for local AI refinement essential for clinical validation.
- Develop a scalable federated research ecosystem enabling equitable AI diagnostics to be compared and delivered across healthcare systems
For further information about the AAIMC-OC Consortium or joining, please contact us at AAIMC_OC@contacts.bham.ac.uk
Current research funding
This work is currently funded by the US Department of Defense Congressionally Directed Medical Research Programmes (CDMRP) FY23 Ovarian Cancer Research Programme Award (OC230126).