Digital Pathology

Digital pathology and AI

Work Package 3

The Cytosponge processing differs from endoscopic biopsy handling and requires quality control measures and expertise in Cytology for reporting. Deep learning methods have been shown to achieve excellent performance on diagnostic tasks, but it is still an open challenge how to optimally combine them with expert knowledge and existing clinical decision pathways. This question is particularly important for the early detection of cancer, where high volume workflows might potentially benefit substantially from automated analysis.

 

In this work package Marcel Gehrung’s team will develop and validate a deep learning framework to analyse samples of the Cytosponge -TFF3 test, a minimally invasive alternative to endoscopy, for detecting Barrett's Esophagus, the main precursor of esophageal cancer. Our approach based on a research prototype exploits screening patterns of expert gastrointestinal pathologists and established decision pathways to define eight triage classes of varying priority for manual expert review.

 

By generating hotspot identification and simultaneous smart report generation our software will be able to substantially reduce the time taken by a pathologist to review Cytosponge samples. This approach will further lay the foundation for tailored, semi-automated decision support systems embedded in clinical workflows with a perspective of using the technology in other diagnostic pathways.

 

Digital pathology developed in WP3 will be used at two points in the implementation pathway – (1) to evaluate the status of cellular samples from the Cytosponge test and (2) to analyse endoscopy samples from patients who have a positive result, for subsequent diagnosis, management and treatment. Dr Hall’s team will assess the associated legal, regulatory and ethical challenges, particularly if the invitation for testing is solely automated in the future. These include consideration of the requirements for information provision, transparency (GDPR Articles 5, 13-15 and Article 22 of the GDPR); evaluating the implications for practice arising from those GDPR obligations including exploring the implications for consent discussions; and producing materials which could be used to support interactions between health professionals and patients. Where relevant, this element may also include consideration of relevant aspects of compliance UK Medical Devices directive and forthcoming national legislation.

© Project DELTA

Cytosponge is a registered trademark of Medtronic