Evaluations of capsule endoscopy software in reducing the reading time and the rate of false negatives by inexperienced endoscopists

Background and objective: Capsule endoscopy (CE) is a comfortable for the patients; however, CE review is time-consuming. The aim of this study was (1) to evaluate the effectiveness of the CE software in reducing the CE reading time and the number of false negatives by beginners, and (2) to determine the learning curve for reading CE images.

Methods: Capsule endoscopic images were captured by Pillcam SB (Given Imaging Ltd, Tokyo, Japan), and analyzed using the proprietary RAPID 5 software. Comparison of CE reading using different software modes: manual mode, automatic mode, and QuickView (QV) mode. Three trainee endoscopists participated as CE readers. Each participant watched CE videos in which positive findings had been predefined by trained endoscopists. Each participant read the same CE record by using one of three different software modes. These were blinded on clinical history of patients. CE reading time was recorded, and the number of false negatives was counted. Each trainee endoscopist read a total of 45 CE videos, in five steps divided into nine videos per step.

Results: There was no significant reader associated difference between the results for the different modes. The QV software did miss some positive findings. Therefore, the total number of instances of FN by the software plus the reader in the QV mode was significantly higher than the others. The reading times in the automatic mode and the QV mode were significantly shorter than that in the manual mode. After the second step, the number of instances of false negatives significantly decreased.

Conclusions: CE software is useful for reducing the reading time. Experience of approximately 20 CE readings can be considered as the first step to becoming an expert.