Forrest Classification for Bleeding Peptic Ulcer: A New Look at the Old Endoscopic Classification

Area of Science:

• Gastroenterology
• Endoscopic imaging
• Medical diagnostics

Background:

• Peptic ulcer bleeding management is complex.
• The Forrest classification aids risk stratification but has variable interpretation.
• Quantitative analysis of endoscopic images for Forrest classification has not been performed.

Purpose of the Study:

• To quantitatively analyze endoscopic images of peptic ulcer bleeding.
• To assess the agreement of Forrest classification among endoscopists.
• To identify distinct visual patterns associated with different Forrest classification stages.

Main Methods:

• Retrieved and reviewed 276 endoscopic still images of peptic ulcer bleeding.
• Assessed intra-rater and inter-rater agreement for Forrest classification.
• Manually delineated ulcer and bleeding areas on images.
• Quantitatively compared image regions of interest across Forrest stages.

Main Results:

• High intra-rater (0.92-0.98) and inter-rater (0.639-0.859) agreement for Forrest classification.
• Higher correlation among tutor and junior endoscopists compared to experienced ones.
• Distinct visual patterns identified for low-risk (Forrest IIC, III) versus high-risk (Forrest I, IIA, IIB) lesions.

Conclusions:

• Good agreement exists in Forrest classification among endoscopists within an institution.
• This study provides the first quantitative analysis and explanation of distinct bleeding ulcer patterns from endoscopic images.
• Findings support the utility of Forrest classification with potential for improved standardization.