Clinical usefulness of a deep learning-based system as the first screening on small-bowel capsule endoscopy reading.

Researchers

Journal

Modalities

Models

Abstract

To examine whether our convolutional neural network (CNN) system based on deep learning could reduce the reading-time of endoscopists without oversight of abnormalities in capsule endoscopy reading process.
Twenty full-videos of small-bowel capsule endoscopy were prepared, each of which included 0-5 lesions of small-bowel mucosal breaks (erosions or ulcerations). At another institute, two reading processes were compared: (A) endoscopist alone readings and (B) endoscopist readings after the first screening by the proposed CNN. In process B, endoscopists read only images picked-up by the CNN. Two experts and four trainees independently read 20 videos each (10 for process A and 10 for process B). Outcomes were the reading-time and the detection rate of mucosal breaks by endoscopists. The gold standard was findings at the original institute by two experts.
The mean reading-time of small-bowel sections by endoscopists was significantly shorter during the process B (expert, 3.1 min; trainee, 5.2 min), compared to the process A (expert, 12.2 min; trainee, 20.7 min) (p < 0.001). For 37 mucosal breaks, the detection rate by endoscopists did not significantly decrease in the process B (expert, 87%; trainee, 55%), compared to the process A (expert, 84%; trainee, 47%). Experts detected all eight large lesions (>5 mm), but trainees could not, even when supported by the CNN.
Our CNN-based system for capsule endoscopy videos could reduce the reading-time of endoscopists, without decreasing the detection rate of mucosal breaks. However, the reading level of endoscopists should be considered on using the system. This article is protected by copyright. All rights reserved.
This article is protected by copyright. All rights reserved.