News, NHS trust

AI tool tested to automate biopsy analysis for coeliac diagnosis in Cambridge

A research programme in Cambridge, funded by the National Institute for Health and Care Research, is supporting Lyzeum Ltd, a spinout from the University of Cambridge, to commercialise its AI tool that aims to accelerate coeliac diagnosis.

Led by Professor Liz Soilleux, honorary consultant pathologist at CUH and professor of diagnostics & biomarkers at the University of Cambridge, the study involved submitting 4,000 sets of biopsy images from five NHS hospitals to the algorithm “to help it distinguish between healthy samples and those with coeliac disease”. When the algorithm was tested on another 650 biopsies, the trust reports it made a correct diagnosis “in 97 out of 100 cases”.

According to CUH, researchers have been working with patient groups, including through Coeliac UK, to “understand the patient response to AI-aided diagnosis”, with a key concern identified around explainability, and difficulties people may experience in understanding what patterns AI is using to come to a conclusion. Ensuring diagnoses are explainable is “key for the researchers”, CUH continues, “and is likely to be a critical step in the AI being approved, and trusted, for use across the NHS”.

Florian Jaeckle, first author on the paper, is quoted as saying that the next step is testing the algorithm “in a much larger clinical sample”, which will place the team “in a position to share this device with the regulator, bringing us nearer to this tool being used in the NHS”.

Digital diagnosis and the role of AI

The UK government recently announced £82.6 million in research funding for three projects, two of which are using emerging technologies such as AI to tackle cancer. The government has also made a commitment to provide researchers with access to “cutting-edge computing resources” to help harness “the power of AI”. An example is PharosAI, a project seeking to bring together NHS and Biobank data onto a “unified, powerful, secure, AI platform”, a joint venture between King’s College London, Queen Mary University of London, Guy’s and St Thomas’ NHS Foundation Trust and Barts Health NHS Trust, aiming to develop AI models to deliver “new breakthroughs” in the diagnosis and treatment of cancer.

A recent news in brief article covered a collaboration between University Hospitals Dorset and Bournemouth University aiming to identify volatile organic compounds (VOCs) specific to skin cancers, and develop a device capable of detecting them.

Elsewhere, Norfolk and Norwich University Hospitals is working with the University of East Anglia to develop a device that helps with identifying the most common causes of dizziness. The Continuous Ambulatory Vestibular Assessment (CAVA) is being tested by patients across the country, analysing hours of eye and head movement data with an aim to help “speed up the diagnosis” of dizziness. 20 hospitals have reportedly been taking part in the clinical trial.