Department of Gastroenterology, Nanjing Drum Tower Hospital, Affiliated Drum Tower Hospital, Medical School of Nanjing University, Nanjing, 210008, Jiangsu, China.
Department of Gastroenterology, Wuxi People's Hospital, Affiliated Wuxi People's Hospital With Nanjing Medical University, Wuxi, 214023, Jiangsu, China.
Surg Endosc. 2022 Oct;36(10):7800-7810. doi: 10.1007/s00464-022-09319-2. Epub 2022 May 31.
Diagnosis of early gastric cancer (EGC) under narrow band imaging endoscopy (NBI) is dependent on expertise and skills. We aimed to elucidate whether artificial intelligence (AI) could diagnose EGC under NBI and evaluate the diagnostic assistance of the AI system.
In this retrospective diagnostic study, 21,785 NBI images and 20 videos from five centers were divided into a training dataset (13,151 images, 810 patients), an internal validation dataset (7057 images, 283 patients), four external validation datasets (1577 images, 147 patients), and a video validation dataset (20 videos, 20 patients). All the images were labeled manually and used to train an AI system using You look only once v3 (YOLOv3). Next, the diagnostic performance of the AI system and endoscopists were compared and the diagnostic assistance of the AI system was assessed. The accuracy, sensitivity, specificity, and AUC were primary outcomes.
The AI system diagnosed EGCs on validation datasets with AUCs of 0.888-0.951 and diagnosed all the EGCs (100.0%) in video dataset. The AI system achieved better diagnostic performance (accuracy, 93.2%, 95% CI, 90.0-94.9%) than senior (85.9%, 95% CI, 84.2-87.4%) and junior (79.5%, 95% CI, 77.8-81.0%) endoscopists. The AI system significantly enhanced the performance of endoscopists in senior (89.4%, 95% CI, 87.9-90.7%) and junior (84.9%, 95% CI, 83.4-86.3%) endoscopists.
The NBI AI system outperformed the endoscopists and exerted potential assistant impact in EGC identification. Prospective validations are needed to evaluate the clinical reinforce of the system in real clinical practice.
窄带成像内镜（NBI）下早期胃癌（EGC）的诊断依赖于专业知识和技能。我们旨在阐明人工智能（AI）是否可以在 NBI 下诊断 EGC，并评估 AI 系统的诊断辅助作用。
在这项回顾性诊断研究中，来自五个中心的 21785 个 NBI 图像和 20 个视频被分为训练数据集（13151 个图像，810 例患者）、内部验证数据集（7057 个图像，283 例患者）、四个外部验证数据集（1577 个图像，147 例患者）和一个视频验证数据集（20 个视频，20 例患者）。所有图像均进行手动标记，并使用 You look only once v3（YOLOv3）对 AI 系统进行训练。接下来，比较了 AI 系统和内镜医生的诊断性能，并评估了 AI 系统的诊断辅助作用。准确性、敏感性、特异性和 AUC 是主要结果。
AI 系统在验证数据集中诊断 EGC 的 AUC 值为 0.888-0.951，并在视频数据集中诊断出所有 EGC（100.0%）。AI 系统的诊断性能优于高级（85.9%，95%CI，84.2-87.4%）和初级（79.5%，95%CI，77.8-81.0%）内镜医生。AI 系统显著提高了高级（89.4%，95%CI，87.9-90.7%）和初级（84.9%，95%CI，83.4-86.3%）内镜医生的诊断性能。
NBI AI 系统优于内镜医生，并在 EGC 识别中具有潜在的辅助作用。需要前瞻性验证来评估该系统在实际临床实践中的临床强化作用。
