AI powered early detection of Barrett’s oesophagus
Artificial intelligence (AI) could help free up pathologist time and allow them to focus on diagnosing the trickiest cases of Barrett’s oesophagus.
Barrett’s oesophagus is a condition where some cells in the oesophagus grow abnormally. Around 3 to 13% of people with Barrett’s oesophagus will develop oesophageal adenocarcinoma, a type of oesophageal cancer, in their lifetime – approximately 11 times higher than in the average person.
It’s thought that many cases of Barrett’s oesophagus go undetected and, because oesophageal cancer is often diagnosed at a late stage, there has been a lot of interest in improving the detection of Barrett’s.
Traditional methods to diagnose Barrett’s oesophagus rely on invasive endoscopies, which are both time-consuming and not available to many patients. To address these challenges, Cambridge researchers have developed innovative tools to enhance early detection.
Professor Rebecca Fitzgerald and her team at the Early Cancer Institute in Cambridge have created the EndoSign—an innovative ‘sponge-on-a-string’ cell-sampling device that allows for non-invasive, efficient sampling, significantly improving the diagnostic process.
Research from the group showed that if EndoSign (previously called Cytosponge) was used to find patients with undiagnosed Barrett’s oesophagus, it can identify ten times more people than the current route. However, if this capsule sponge becomes more routinely used in the NHS, this could increase the demand on the already busy NHS care pathway.
The Markowetz Group has developed artificial intelligence tools to automate the analysis of samples collected by the EdnoSign, reducing the workload for pathologists by 57%, without compromising the quality of care.
Using the biomarker TFF3, which is a hallmark of Barrett’s oesophagus, the researchers trained the algorithm to recognise overexpression of this protein in goblet cells, which are found in the oesophagus.
Working with pathologists, they developed criteria that were able to distinguish between signals that indicated presence of Barrett’s oesophagus from the noise of other cell populations.
The development of these diagnostic tools has led to Cyted, a spin-out company founded by CRUK CI alumnus Marcel Gehrung, based on his PhD work in the Markowetz lab.
The deep-learning tools developed by the team are set to be used in the upcoming Barrett’s Esophagus Screening Trial (BEST4), aiming to further validate and implement these innovations in clinical practice.
