Web-based system for training and dissemination of a magnification chromoendoscopy classification

Area of Science:

• Gastroenterology
• Medical Education
• Health Informatics

Background:

• Magnification chromoendoscopy (MC) is vital for gastric lesion classification.
• Standardizing MC interpretation and assessing observer learning curves can be challenging.
• Web-based technologies offer potential solutions for training and evaluation.

Purpose of the Study:

• To evaluate web-based technologies for assessing the learning curve and reproducibility of a simplified gastric MC classification.
• To determine the effectiveness of a hybrid web-based system for training endoscopists.

Main Methods:

• A multicenter trial utilized a hybrid approach with CD-ROM and a remote online questionnaire.
• Three endoscopists independently classified 10 randomly selected MC films, with feedback provided.
• Data were collected centrally to track observer agreement and learning progression.

Main Results:

• Both intra- and inter-observer agreement significantly improved with repeated observations, reaching excellent levels (kappa = 0.94).
• Observer agreement with the reference standard also increased, with 100% validity achieved by the 4th observation.
• Consideration of a 4th subgroup showed near-perfect inter-observer agreement (kappa = 0.92) and improved sensitivity.

Conclusions:

• The simplified MC classification is easily learnable and reproducible, evidenced by high observer agreement.
• The developed web system is a valuable tool for defining, educating, and disseminating knowledge in endoscopic image-based diagnostics.
• Web-based platforms can enhance training and standardization in various image-based medical fields.