Search

Background and Aims: Computer-aided detection software (CADe) has shown promising results in real-time polyp detection, but a limited head-to-head comparison of the available CADe systems has been performed. Moreover, such systems have not been compared to endoscopists using standardized videos. This study aims to compare the performance of three CADe systems in detecting polyps, employing a novel standardized methodology., Methods: Videos from 300 colonoscopies conducted at Oslo University Hospital were analyzed. Short video clips (20-45 s) presenting normal mucosa or polyps were randomly selected. These videos were then streamed through each CADe system from Medtronic, Olympus, and Augere Medical. Each system featured diverse configurations, resulting in a total of six software settings. Sensitivity and false positivity (FP) were assessed by comparing the CADe systems to both the mean of the systems and pairwise between them. Furthermore, the systems' performance was compared to the performance of five endoscopists., Results: CADe systems' sensitivity ranged between 84.9% and 98.7%, with statistically significant differences observed between the systems, both in comparison to the mean and to each other. FP rates ranged between 1.2% and 5.6%, also differing statistically significantly between the systems. The CADe systems achieving the highest sensitivity also exhibited the highest FP. Statistically significant differences in the alert delay were observed between different CADe systems and endoscopists., Conclusions: This study highlights significant differences between commercially available CADe software regarding sensitivity and FP, but a superior performance compared to endoscopists. The software with the highest sensitivity also exhibited the highest FP, highlighting the need for further refinement., Competing Interests: Authors Pia Helén Smedsrud, Tor Jan D. Berstad, and Thomas de Lange are employed and shareholders in Augere Medical. Håvard Espeland, Andreas Petlund, Pål Halvorsen, and Michael A. Riegler are shareholders in Augere Medical., (© 2025 The Author(s). DEN Open published by John Wiley & Sons Australia, Ltd on behalf of Japan Gastroenterological Endoscopy Society.)

Artificial intelligence (AI) is predicted to have profound effects on the future of video capsule endoscopy (VCE) technology. The potential lies in improving anomaly detection while reducing manual labour. Existing work demonstrates the promising benefits of AI-based computer-assisted diagnosis systems for VCE. They also show great potential for improvements to achieve even better results. Also, medical data is often sparse and unavailable to the research community, and qualified medical personnel rarely have time for the tedious labelling work. We present Kvasir-Capsule, a large VCE dataset collected from examinations at a Norwegian Hospital. Kvasir-Capsule consists of 117 videos which can be used to extract a total of 4,741,504 image frames. We have labelled and medically verified 47,238 frames with a bounding box around findings from 14 different classes. In addition to these labelled images, there are 4,694,266 unlabelled frames included in the dataset. The Kvasir-Capsule dataset can play a valuable role in developing better algorithms in order to reach true potential of VCE technology.