Exploring the potential of AI-enabled medical imaging

Cancer survival in the UK lags behind other developed nations and a significant percentage (21%) of cancers present as emergencies. Yet early detection of cancer, which may pick up smaller tumours, can improve treatment options and reduce disease mortality.

AI-enabled medical imaging has the potential to enable faster, more accurate diagnosis and surveillance. But the use of AI is still in its early stages and challenges remain regarding governance, ethical treatment of patient information, linking patient data in multiple locations, and the accuracy of AI-generated results.

ESPRC IAA (Partnership) funding enabled the development of a novel academic-hospital-industry partnership to explore how AI tools might be integrated into clinical practice in the assessment of upper gastrointestinal (GI) cancers. The project ran from August 2021 to March 2022 and partners included the Department of Engineering Science, Oxford University Hospitals (OUH), and leading global medical technology company GE HealthCare (GEHC).

Partners held regular meetings to understand the analysis needs of the clinical context and the potential for AI solutions at every stage (primary healthcare, specialist clinics etc) of the cancer pathway. The meetings, which involved researchers, analysts, clinicians, data management experts, and professional software engineers, informed technical work to develop and apply AI tools to real world/time patient datasets.

A significant outcome of the project was the developed of the concept of AI augmented reports for endoscopy and pathology and how these could be integrated into the clinical workflow. Here, the experience of GEHC’s OncoFlow platform has been critical. Based on these learnings the group is now working on ways to test AI algorithms in the clinical setting. Other significant project outcomes include:

• Better understanding of the complexities and ethical considerations of sharing patient data through AI tools.
• Better understanding for clinical and industry partners of how AI platforms can support real-world clinical pathways.
• Gathering of preliminary data which has helped leverage significant funding for further research.

Professor Jens Rittscher, Department of Engineering Science, comments: "The potential for AI-enabled medical imaging to transform diagnosis and management of cancer is very exciting. As scientists we will continue to work with industry and clinicians to bring this closer to the stage where it can really improve patient outcomes".

As part of the NIHR Biomedical Research Centre Imaging Theme the groups is now working on concrete projects to explore the application of AI imaging tools in the clinical setting.

Further information

OU-OUH-GE IAA Partnership project on data-driven healthcare technologies for oncology was funded by the EPSRC Impact Acceleration Account Partnership fund from August 2021 – March 2022

Project partners

Oxford University Department of Engineering Science

Oxford University Hospitals

GE HealthCare