Validated computed cleansing score for video capsule endoscopy

Background and Aim: Diagnostic yield of video capsule endoscopy may be hampered by intestinal content or air bubbles. A major limitation in video capsule-related study is the lack of a validated objective score for bowel preparation quality. We aimed to design and validate a computed small bowel preparation score for research and clinical use. Methods: Two experienced physicians reached a consensus regarding bowel preparation quality based on known criteria used in previous studies and their confidence of an accurate medical interpretation of the procedure. A computed algorithm based on the pixels in the color bar was created and validated. Concordance between the gastroenterologists’ agreement (gold standard) and the computed analysis was assessed. Results: Of 85 videos studied, 44 (52%), 13 (15%) and 28 (33%) had adequate, borderline and inadequate bowel preparation, respectively, according to the gastroenterologists’ agreement. Computer analysis restricted to adequate and inadequate cases yielded accurate classification of bowel preparation in 65/72 cases (90% agreement, sensitivity 95%, specificity 82%, total accuracy 90%, Kappa 0.79). When adding the borderline definition, the computer analysis correctly classified 71/85 of the cases, yielding an overall agreement of 84% (Kappa 0.72). Minute-by-minute analysis of 10 cases also yielded an agreement of 91.4%. Conclusion: The present study introduces a user-friendly computer analysis-based small bowel preparation score, which demonstrated excellent concordance with the physician's assessment. This score holds promise as a standardization tool in research and clinical practice of video capsule endoscopy. Further validation is warranted.