REVOLUTIONISING PATHOLOGY WITH ARTIFICIAL INTELLIGENCE
The landscape of medical diagnostics is on the cusp of a transformative change with the advent of innovative Artificial Intelligence (AI) models. A team of experts from the School of Cancer Sciences have developed a revolutionary AI model that turns pathology images into a self-discovered language of cancer and corresponding paragraphs of AI text.
The model interprets those paragraphs to understand the context and relationship of the tissue. This is an unprecedented approach to image analysis in medicine and unlocks the training of large AI models without human labels, leading to a new generation of clinical tools.
One of the most impactful tools that the model will power is a pan-cancer triage tool. On average, only 73% of people start treatment within the NHS target time of 62 days, and for some cancers it can be as low as 49%. This is leading to poor patient outcomes and having an untold cost in lives and knock-on effects in healthcare economics. Using this new approach, it is possible to rapidly and automatically detect malignancies, reduce time until treatment, improve patient outcomes and mitigate the waiting time crisis in the NHS. Another impactful area where it is being applied is to help discover new targets to assist drug discovery and development.
contact: innovation@glasgow.ac.uk
THE TEAM
This innovation has evolved within the School of Cancer Sciences. The team behind this innovation brings a wealth of expertise to the table, including expertise in AI, cancer research, and digital pathology. They are joined by an expert in biomarker discovery and a practising pancreatic cancer surgeon, in addition to a molecular pathologist with a strong background in biomarker discovery and development.
MEDTECH INNOVATION FUND SUPPORT
The funding has enabled the team to scale the model, applying the technology to 100,000 clinical whole slide images, working with the NHS and Microsoft to achieve this objective. The process to access real clinical data on which to validate the model has been a critical de-risking exercise, which would not have been possible without MedTech funding support. This step is critical as it provides tangible proof of the model’s efficacy and commercial viability.
FINAL OVERVIEW
The innovative AI technology developed by the School of Cancer Sciences represents a transformative leap forward in pathology, promising to revolutionise cancer diagnostics and patient care. By enabling rapid, accurate, and self-supervised interpretation of pathology images, this model is set to significantly reduce diagnostic waiting times, improve patient outcomes, and ease NHS pressures. With the critical support from the MedTech Innovation Fund and strategic partnerships, the team’s pioneering work brings us closer to a future where AI-powered pathology is standard practice, ultimately saving lives and advancing medical discovery.