Original articleClinical endoscopyComputer-aided detection of early Barrett’s neoplasia using volumetric laser endomicroscopy
Section snippets
Patients and image database
The VLE images used in this study were derived from a database of high-quality 1-to-1 VLE-histology correlations. The construction of this database has been described previously.14 In short, endoscopic resection specimens from patients with BE with and without early neoplasia were used. The fresh specimens were scanned ex vivo with VLE in a custom-made fixture. Guided by previously placed in vivo and ex vivo markers, histologic sectioning was performed. Histology slides and VLE frames were
Patients and VLE-histology database
The previously constructed VLE-histology database consisted of 52 endoscopic resection specimens from 29 BE patients (mean age, 67 years [SD ±8.4]; 22 men). The worst histologic diagnosis per patient was NDBE in 6, low-grade dysplasia in 2, HGD in 5, and EAC in 16. In total, 86 VLE-histology matches containing 200 areas of interest were constructed. For this study 30 NDBE and 30 HGD/EAC areas of interest of sufficient quality were selected. These 60 images were derived from 19 different
Discussion
Our results show that a computer algorithm based on clinically derived features is capable of identifying early BE neoplasia on ex vivo VLE images. In fact, compared with VLE experts who scored the same ex vivo VLE images, the algorithm performed even better in identifying neoplasia (AUC, .89-.95 vs .81 for experts). We envision that an objective and quantitative interpretation of VLE by a computer-aided algorithm, not hindered by interobserver variability, will be an important tool to aid in
References (22)
- et al.
Comprehensive microscopy of the esophagus in human patients with optical frequency domain imaging
Gastrointest Endosc
(2008) - et al.
How good are experts in identifying early Barrett’s neoplasia in endoscopic resection specimens using volumetric laser endomicroscopy?
Gastroenterology
(2016) - et al.
How good are experts in identifying endoscopically visible early Barrett’s neoplasia on in vivo volumetric laser endomicroscopy [abstract]?
Gastrointest Endosc
(2016) - et al.
Comprehensive esophageal microscopy by using optical frequency-domain imaging (with video)
Gastrointest Endosc
(2007) - et al.
A new paradigm shift in endoscopy: from interpretation to automated image analysis?
Gastrointest Endosc
(2016) - et al.
Esophageal adenocarcinoma incidence: Are we reaching the peak?
Cancer Epidemiol Biomarkers Prev
(2010) - et al.
British Society of Gastroenterology guidelines on the diagnosis and management of Barrett’s oesophagus
Gut
(2014) - et al.
ACG Clinical Guideline: diagnosis and management of Barrett’s esophagus
Am J Gastroenterol
(2015) - et al.
Safety and feasibility of volumetric laser endomicroscopy in patients with Barrett’s esophagus (with videos)
Gastrointest Endosc
(2015) - et al.
Identification of volumetric laser endomicroscopy features predictive for early neoplasia in Barrett’s esophagus using high-quality histological correlation
Gastrointest Endosc
(2017)
Comparative diagnostic performance of volumetric laser endomicroscopy and confocal laser endomicroscopy in the detection of dysplasia associated with Barrett’s esophagus
Gastrointest Endosc
Cited by (107)
Feasibility and Safety of Tethered Capsule Endomicroscopy in Patients With Barrett's Esophagus in a Multi-Center Study
2022, Clinical Gastroenterology and HepatologyArtificial Intelligence and the Future of Gastroenterology and Hepatology
2022, Gastro Hep AdvancesAdvanced endoscopic imaging for detection of Barrett's esophagus
2024, Clinical EndoscopyResearch progress of artificial intelligence in esophageal diseases
2023, Chinese Journal of Digestive Endoscopy
DISCLOSURE: The following author disclosed financial relationships relevant to this publication: J. J. Bergman: Research support recipient from Ninepoint Medical, Olympus Endoscopy, Erbe, and Fujifilm; consultant for Olympus Endoscopy and Fujifilm. All other authors disclosed no financial relationships relevant to this publication. Research support for this study was provided by an unrestricted grant from NinePoint Medical Inc.
See CME section; p. 903.