Department of Gastroenterology and Hepatology, University Hospitals Leuven, Herestraat 49, 3000, Leuven, Belgium; Department of Translational Research in Gastrointestinal Diseases (TARGID), Catholic University Leuven, Herestraat 49, 3000, Leuven, Belgium.
Best Pract Res Clin Gastroenterol. 2021 Jun-Aug;52-53:101722. doi: 10.1016/j.bpg.2020.101722. Epub 2020 Dec 29.
Artificial intelligence (AI) is of keen interest for global health development as potential support for current human shortcomings. Gastrointestinal (GI) endoscopy is an excellent substrate for AI, since it holds the genuine potential to improve quality in GI endoscopy and overall patient care by improving detection and diagnosis guiding the endoscopists in performing endoscopy to the highest quality standards. The possibility of large data acquisitioning to refine algorithms makes implementation of AI into daily practice a potential reality. With the start of a new era adopting deep learning, large amounts of data can easily be processed, resulting in better diagnostic performances. In the upper gastrointestinal tract, research currently focusses on the detection and characterization of neoplasia, including Barrett's, squamous cell and gastric carcinoma, with an increasing amount of AI studies demonstrating the potential and benefit of AI-augmented endoscopy. Deep learning applied to small bowel video capsule endoscopy also appears to enhance pathology detection and reduce capsule reading time. In the colon, multiple prospective trials including five randomized trials, showed a consistent improvement in polyp and adenoma detection rates, one of the main quality indicators in endoscopy. There are however potential additional roles for AI to assist in quality improvement of endoscopic procedures, training and therapeutic decision making. Further large-scale, multicenter validation trials are required before AI-augmented diagnostic gastrointestinal endoscopy can be integrated into our routine clinical practice.
人工智能（AI）在全球健康发展中备受关注，因为它有可能为人类目前的不足提供支持。胃肠内镜是 AI 的极佳载体，因为它具有真正的潜力，可以通过提高检测和诊断水平，从而改善胃肠内镜的质量和整体患者护理水平，指导内镜医生达到最高质量标准。大量获取数据以完善算法的可能性使得 AI 实施到日常实践中成为一种潜在的现实。随着深度学习新时代的开始，大量数据可以轻松处理，从而提高诊断性能。在上消化道，研究目前集中在肿瘤的检测和特征描述上，包括 Barrett 食管、鳞状细胞癌和胃癌，越来越多的 AI 研究证明了 AI 增强内镜的潜力和益处。深度学习应用于小肠视频胶囊内镜似乎也增强了病理学检测并减少了胶囊阅读时间。在结肠中，多项前瞻性试验，包括五项随机试验，一致显示出息肉和腺瘤检测率的提高，这是内镜质量的主要指标之一。然而，人工智能在协助提高内镜手术的质量、培训和治疗决策方面可能还有其他潜在作用。在 AI 增强的诊断性胃肠内镜能够整合到我们的常规临床实践之前，还需要进行更多的大规模、多中心验证试验。
