Advanced imaging techniques for detection of Barrett’s neoplasia

Patients with Barrett’s esophagus, in which the normal lining of the esophageal wall has been replaced under influence of chronic acid reflux, have an increased risk of developing esophageal adenocarcinoma. Barrett’s patients therefore undergo regular endoscopic surveillance. When esophageal cancer is detected in an early stage it can be treated endoscopically with an excellent prognosis. However, early cancer has a subtle appearance and can therefore be missed.
To increase neoplasia detection in Barrett’s esophagus (BE), many advanced endoscopic imaging techniques have been developed over the past decades. Several of these techniques have been discussed in this thesis, including optical chromoscopy, magnification endoscopy and volumetric laser endomicroscopy (VLE). In this thesis, we have investigated the application of these novel imaging techniques for detection and characterization of BE neoplasia. We described three major challenges of Barrett surveillance: 1) a subjective interpretation of endoscopic imagery by endoscopists, 2) the increase of available information within a Barrett segment, and 3) sampling error associated with random biopsies. To address these challenges, application of artificial intelligence (AI) was explored for enhanced endoscopic image interpretation. Additionally, VLE was studied, which offers the possibility to visualize the different esophageal wall layers and thereby guide targeted biopsies.
Finally, we developed an AI-tool to aid in the interpretation of the complex information within VLE imagery in order to improve BE neoplasia detection. Future studies should evaluate our AI-tools during live clinical procedures.