Tada Tomohiro Institute of Gastroenterology and Proctology, Saitama, Japan; Department of Surgery, Teikyo University School of Medicine, Tokyo, Japan.
Tada Tomohiro Institute of Gastroenterology and Proctology, Saitama, Japan; Department of Surgery, Sanno Hospital, The International University of Health and Welfare, Tokyo, Japan; Department of Surgical Oncology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan.
Gastrointest Endosc. 2019 Feb;89(2):416-421.e1. doi: 10.1016/j.gie.2018.10.020. Epub 2018 Oct 24.
Evaluation of endoscopic disease activity for patients with ulcerative colitis (UC) is important when determining the treatment of choice. However, endoscopists require a certain period of training to evaluate the activity of inflammation properly, and interobserver variability exists. Therefore, we constructed a computer-assisted diagnosis (CAD) system using a convolutional neural network (CNN) and evaluated its performance using a large dataset of endoscopic images from patients with UC.
A CNN-based CAD system was constructed based on GoogLeNet architecture. The CNN was trained using 26,304 colonoscopy images from a cumulative total of 841 patients with UC, which were tagged with anatomic locations and Mayo endoscopic scores. The performance of the CNN in identifying normal mucosa (Mayo 0) and mucosal healing state (Mayo 0-1) was evaluated in an independent test set of 3981 images from 114 patients with UC, by calculating the areas under the receiver operating characteristic curves (AUROCs). In addition, AUROCs in the right side of the colon, left side of the colon, and rectum were evaluated.
The CNN-based CAD system showed a high level of performance with AUROCs of 0.86 and 0.98 to identify Mayo 0 and 0-1, respectively. The performance of the CNN was better for the rectum than for the right side and left side of the colon when identifying Mayo 0 (AUROC = 0.92, 0.83, and 0.83, respectively).
The performance of the CNN-based CAD system was robust when used to identify endoscopic inflammation severity in patients with UC, highlighting its promising role in supporting less-experienced endoscopists and reducing interobserver variability.
评估溃疡性结肠炎（UC）患者的内镜疾病活动度对于确定治疗选择非常重要。然而，内镜医师需要一定的培训期才能正确评估炎症活动度，并且存在观察者间的变异性。因此，我们使用卷积神经网络（CNN）构建了一个计算机辅助诊断（CAD）系统，并使用来自 UC 患者的大量内镜图像数据集评估了其性能。
基于 GoogLeNet 架构构建了基于 CNN 的 CAD 系统。该 CNN 使用来自总共 841 例 UC 患者的 26,304 张结肠镜图像进行了训练，这些图像被标记为解剖位置和 Mayo 内镜评分。在一个来自 114 例 UC 患者的 3981 张图像的独立测试集中，通过计算受试者工作特征曲线下的面积（AUROCs），评估了 CNN 识别正常黏膜（Mayo 0）和黏膜愈合状态（Mayo 0-1）的性能。此外，还评估了右半结肠、左半结肠和直肠的 AUROCs。
基于 CNN 的 CAD 系统表现出较高的性能，分别为 0.86 和 0.98 的 AUROC 来识别 Mayo 0 和 0-1。在识别 Mayo 0 时，CNN 的性能在直肠优于右半结肠和左半结肠（AUROC 分别为 0.92、0.83 和 0.83）。
基于 CNN 的 CAD 系统在识别 UC 患者的内镜炎症严重程度方面表现稳健，这突出了其在支持经验较少的内镜医师和减少观察者间变异性方面的有前途的作用。
