Cookies

Like most websites The Pathologist uses cookies. In order to deliver a personalized, responsive service and to improve the site, we remember and store information about how you use it. Learn more.
Diagnostics Digital and computational pathology, Oncology, Software and hardware

Deep Learning Diagnostics for Barrett’s Esophagus

With Barrett’s esophagus patients at a significantly higher risk of developing esophageal adenocarcinoma than the average person, the need for early diagnosis and monitoring is vital (1). To meet this need, researchers have developed and trained a deep learning algorithm to analyze samples from a Cytosponge-TFF3 test (2) – a device that collects cells from the lining of the esophagus, allowing patient triage for endoscopy. “The pathology analysis workflow for Cytosponge is laborious, but it consists of repetitive elements that can be automated using computational pathology to find cases where the automated analysis breaks down and human experts are required,” says lead researcher Marcel Gehrung.

The system was trained to triage samples based on a two-variable, two-step process. “We defined two different scores (quality and diagnosis) and divided them into three tiers (no/low/high confidence),” explains Gehrung. “For combinations of no/low confidence between quality and diagnosis, we then decide that the sample should be analyzed by a human expert, rather than an automated algorithm.”

The tool has clear benefits for laboratory medicine professionals, not least in reducing pathologists’ workload by 57 percent while maintaining diagnostic standards (2). “It enables pathologists to focus more time on difficult cases – therefore reducing error rates due to increasing workload,” says Gehrung.  “It can also help to build more trust in (semi-)automated analyses because the ‘easy’ cases are those in which automated performance can best demonstrate its use.”

But that’s not all – it can also fit seamlessly into the existing laboratory workflow. “A tool like this can be used on whole-slide images immediately after scanning,” says Gehrung. “The algorithm then decides whether a human is required or its own assessment is confident enough. If a human needs to review the case, then it follows the normal analysis pathway – but if the algorithm produces the report, then no human involvement is required.”

Though the algorithm won’t replace pathologists’ valuable role in the patient pathway, it could help them deliver the best possible service for Barrett’s esophagus patients whose risk of esophageal cancer might otherwise go undetected.

Receive content, products, events as well as relevant industry updates from The Pathologist and its sponsors.

When you click “Subscribe” we will email you a link, which you must click to verify the email address above and activate your subscription. If you do not receive this email, please contact us at [email protected].
If you wish to unsubscribe, you can update your preferences at any point.

  1. P Gatenby et al., World J Gastroenterol, 20, 9611 (2014). PMID: 25071359.
  2. M Gehrung et al., Nat Med, [Online ahead of print] (2021). PMID: 33859411.
About the Author
Olivia Gaskill

During my undergraduate degree in psychology and Master’s in neuroimaging for clinical and cognitive neuroscience, I realized the tasks my classmates found tedious – writing essays, editing, proofreading – were the ones that gave me the greatest satisfaction. I quickly gathered that rambling on about science in the bar wasn’t exactly riveting for my non-scientist friends, so my thoughts turned to a career in science writing. At Texere, I get to craft science into stories, interact with international experts, and engage with readers who love science just as much as I do.

Register to The Pathologist

Register to access our FREE online portfolio, request the magazine in print and manage your preferences.

You will benefit from:
  • Unlimited access to ALL articles
  • News, interviews & opinions from leading industry experts
  • Receive print (and PDF) copies of The Pathologist magazine

Register