Quantitative analysis of red color sign in the endoscopic evaluation of esophageal varices

Bleeding due to esophageal variceal rupture is associated with an extremely high mortality rate. Variceal bleeding is frequent in patients who have a red color sign on endoscopy. However, the red color sign is subjectively evaluated on the basis of color tone and the shape of the varices. To allow standardization and facilitate consensus, an objective method of assessing the red color sign is needed. In this study, a system was established for quantifying the red color sign during endoscopic evaluation.

Between July 1995 and February 1997, 55 untreated patients with portal hypertension and esophageal varices identified on upper gastrointestinal endoscopy were enrolled in the study. Images obtained about 5 cm oral to the esophagogastric junction during endoscopy were stored on magnetic optical disks using an endoscopic image processor. The still images were transmitted to a computer and analyzed using computer software. The RGB components (R, red; G, green; B, blue) were measured at points showing flare consistent with the red color sign. The endoscopic assessment was based on the Japanese Research Society for Portal Hypertension's general rules for recording endoscopic findings in esophagogastric varices.

The ratio of the red color area to the variceal area increased with increasing red color grade. There were significant positive correlations between the R and G, and G and B components. This suggests that comparing the R components alone would allow assessment of the color differences in the red color area and in the varices. The R value was significantly higher in the red color area (115 +/- 20) than in the varices (57 +/- 19). An R value of 90 was found at the boundary between the two parts (P < 0.001).

The red color area can be automatically calculated and quantified using the analysis program. Improvements in data storage methods may allow real-time evaluation during endoscopy in the future.