March 27, 2025
AI (artificial intelligence) tools developed by researchers at NIHR Cambridge Biomedical Campus can dramatically increase the diagnosis speed of celiac disease patients. Automating biopsy image analysis allows samples to be processed faster, reduced waiting lists, and enable specialized pathologists to help diagnose and treat more people.
Celiac disease affects approximately 1 in 100 people who experience symptoms such as stomach cramps, diarrhea, skin rash, weight loss, fatigue, and anemia as a result of eating gluten.
The diagnosis of celiac disease is made by a specially trained pathologist who analyzes biopsy samples taken from the intestine. This can be a time-consuming process and it is not always possible to achieve decisive results.
“Everything that makes the system faster must be good because if you know you’re diagnosed and you can’t have gluten, you know what to do.”
Liz Cox, 80, lives with celiac disease
A study published today (March 27, 2025) in the New England Journal of Medicine demonstrates how AI can accelerate this process and provide comparable levels of accuracy while reducing the pressure on healthcare resources. It can also provide solutions in developing countries with little or no access to specialized pathologists.
Although AI has already been widely investigated for its ability to detect cancer cells in biopsies, this is the first time that the same has been attempted with celiac disease.
“Celiac disease can affect one in 100 people and cause serious illnesses, but getting a diagnosis is not easy. It can take years to get an accurate diagnosis. These delays can last at the time of strong pressure on the healthcare system.
Professor Liz Soilleux, Research Lead and Honorary Consultant Pathologist
In this study, we showed 4,000 sets of biopsy images collected from five NHS hospitals that a type of AI known as machine learning algorithms helps distinguish between healthy and celiac disease samples.
When AI was tested with another 650 biopsies, it made the correct diagnosis in 97 out of 100 cases when compared to the diagnosis made by human pathologists. This is a groundbreaking achievement as previous work by the same team shows that human pathologists often disagree with diagnosis.
“Whether AI suffers from celiac disease as well as experienced pathologists, it is the first time that AI is diagnosed in the same way as experienced pathologists. As we trained on datasets generated under different conditions, biopsies should be able to work in a wide range of settings where biopsies are processed differently and imaged.
“The next step is to test the algorithm on a much larger clinical sample, put this device in a position to share with regulators, bringing us closer to this tool used by the NHS.”
Dr. Florian Jeckle in the Department of Pathology and first author of the paper
The study is led by Professor Liz Soilleux, an honorary consultant pathologist at Cambridge University Hospital NHS Foundation Trust (CUH) and professor of diagnostics and biomarkers at Cambridge University.
Learn how to participate in research at CUH with Love Research
Researchers work with patient groups through Celiac UK to understand patients’ responses to A-Aided diagnosis.
The key concern that they identified as being addressed by patients and clinicians is “explanability.” AI can reach conclusions based on the patterns seen in the data displayed, but at this point it is not always easy for humans to understand what those patterns are or to see if they are actually related to the disease being diagnosed.
Creating an explanatory AI diagnosis is key for researchers and could be a critical step in AI approved and trusted for use across the NHS.
This study was funded by the National Institute of Health and Therapy. Lyzeum Ltd is a spin-out company at Cambridge University and is set up to commercialize this AI tool.
