Abstract:Background: Digital single-operator cholangioscopy (DSOC) (SpyGlass DS™, Boston Scientific, MA, USA) allows for high-definition imaging of the biliary tree. The superior visualization has led to the development of two different sets of criteria to evaluate and classify indeterminate biliary strictures: the Monaco criteria and the criteria in Carlos Robles-Medranda's publication (CRM). Our objective was to assess the interrater agreement (IA) of DSOC interpretation for indeterminate biliary strictures using the two newly published criteria. Methods: Forty de-identified DSOC video recordings were sent to 15 interventional endoscopists with experience in cholangioscopy. They were asked to score the videos based on the presence of Monaco Classification criteria: stricture, lesion, mucosal changes, papillary projections, ulceration, white linear bands or rings, and vessels. Next, they scored the videos using CRM criteria: villous pattern, polypoid pattern, inflammatory pattern, flat pattern, ulcerate pattern and honeycomb pattern. The endoscopists then diagnosed the recordings as neoplastic or non-neoplastic based on the criteria. Intraclass correlation (ICC) analysis was done to evaluate interrater agreement for both criteria set and final diagnosis. Results: Recordings of 26 malignant lesions and 14 benign lesions were scored. The IA using both the Monaco criteria and CRM criteria ranged from poor to excellent (range 0.1-0.76) and (range 0.1-0.62), respectively. Within the Monaco criteria, IA was excellent for lesion (0.75) and fingerlike papillary projections (0.74); good for tortuous vessels (0.7), mucosal features (0.62), uniform papillary projections (0.53), and ulceration (0.58); and fair for white linear bands (0.4). Within the CRM criteria, the IA was good for villous pattern (0.62), flat pattern (0.62), and honeycomb pattern; fair for ulcerated pattern (0.56), polypoid pattern (0.52) and inflammatory pattern (0.54). The diagnostic IA using Monaco criteria was good (0.65), while the diagnostic IA using CRM was fair (0.58). The overall diagnostic accuracy using the Monaco classification was 61% and CRM criteria were 57%. Conclusion: The IOA and accuracy rate of DSOC using visual criteria from both Monaco Criteria and CRM are similar. However, some criteria from both sets suffer from poor IA, thus affecting the overall diagnostic accuracy. More formal training and refinements in visual criteria with additional validation are needed to improve diagnostic accuracy. Trial registration: ClinicalTrials.gov Identifier: NCT02166099.

Identification of patients with malignant biliary strictures using a cholangioscopy-based deep learning artificial intelligence (with video)

Artificial intelligence in endoscopic imaging for detection of malignant biliary strictures and cholangiocarcinoma: a systematic review

Multimodal Convolutional Neural Network-Based Algorithm for Real-Time Detection and Differentiation of Malignant and Inflammatory Biliary Strictures in Cholangioscopy: a Proof of Concept Study with Video

Cholangioscopy-based convoluted neuronal network vs. confocal laser endomicroscopy in identification of neoplastic biliary strictures

A Real-Time Interpretable Artificial Intelligence Model for the Cholangioscopic Diagnosis of Malignant Biliary Stricture (with Videos).

A Real-Time Interpretable Ai Model for the Cholangioscopic Diagnosis of Malignant Biliary Stricture

A real-time interpretable artificial intelligence model for the cholangioscopic diagnosis of malignant biliary stricture

Deep Learning for Accurate Diagnosis of Liver Tumor Based on Magnetic Resonance Imaging and Clinical Data

Convolutional neural network for the diagnosis of early gastric cancer based on magnifying narrow band imaging

A new artificial intelligence system successfully detects and localises early neoplasia in Barrett's esophagus by using convolutional neural networks

Using neural networks to autonomously assess adequacy in intraoperative cholangiograms

Tu1075 DEEP CONVOLUTIONAL NEURAL NETWORKS FOR RECOGNITION OF ATROPHIC GASTRITIS AND INTESTINAL METAPLASIA BASED ON ENDOSCOPY IMAGES

Deep Learning in Endoscopic Ultrasound: A Breakthrough in Detecting Distal Cholangiocarcinoma

Convolutional neural network-based artificial intelligence for the diagnosis of early esophageal cancer based on endoscopic images: A meta-analysis

Artificial Intelligence in Medical Imaging for Cholangiocarcinoma Diagnosis: A Systematic Review with Scientometric Analysis

Accuracy of convolutional neural network-based artificial intelligence in diagnosis of gastrointestinal lesions based on endoscopic images: A systematic review and meta-analysis

A deep learning-based system for identifying differentiation status and delineating the margins of early gastric cancer in magnifying narrow-band imaging endoscopy

Artificial Intelligence in Gastric Cancer: Identifying Gastric Cancer Using Endoscopic Images with Convolutional Neural Network

Diagnosis of Gastric Lesions Through a Deep Convolutional Neural Network.

Real-time automated diagnosis of precancerous lesions and early esophageal squamous cell carcinoma using a deep learning model (with videos)

Digital Cholangioscopic Interpretation: When North Meets the South