Copyright
©The Author(s) 2025. Published by Baishideng Publishing Group Inc. All rights reserved.
World J Gastrointest Endosc. Oct 16, 2025; 17(10): 113184
Published online Oct 16, 2025. doi: 10.4253/wjge.v17.i10.113184
Published online Oct 16, 2025. doi: 10.4253/wjge.v17.i10.113184
Deep learning meets small-bowel capsule endoscopy: A step toward faster and more consistent diagnosis of obscure gastrointestinal bleeding
Heng-Rui Liu, Cancer Research Institute, Jinan University, Guangzhou 510632, Guangdong Province, China
Author contributions: Liu HR finished the paper.
Conflict-of-interest statement: The author declare that has no conflict of interest.
Open Access: This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: https://creativecommons.org/Licenses/by-nc/4.0/
Corresponding author: Heng-Rui Liu, Researcher, Cancer Research Institute, Jinan University, No. 601 Huangpu West Avenue, Tianhe District, Guangzhou 510632, Guangdong Province, China. lh@yinuobiomedical.cn
Received: August 18, 2025
Revised: August 25, 2025
Accepted: September 22, 2025
Published online: October 16, 2025
Processing time: 59 Days and 10.2 Hours
Revised: August 25, 2025
Accepted: September 22, 2025
Published online: October 16, 2025
Processing time: 59 Days and 10.2 Hours
Core Tip
Core Tip: Small-bowel capsule endoscopy is invaluable for evaluating obscure gastrointestinal bleeding but remains limited by lengthy review times and interobserver variability. This study introduces a deep learning system that integrates gastrointestinal localization with multi-lesion detection in full-length capsule videos, achieving high accuracy and cutting reading time from nearly an hour to just minutes. By closely mimicking human reading workflow and validating across centers, the work highlights a practical step toward clinically deployable artificial intelligence assistance, with the potential to standardize interpretation and improve efficiency in routine practice.