Department of Gastroenterology, Nanfang Hospital, Southern Medical University, Guangzhou, China.
Mechanobiology Institute, National University of Singapore, Singapore; Institute of Bioengineering and Nanotechnology, Agency for Science, Technology and Research (A∗STAR), Singapore.
Gastrointest Endosc. 2021 Sep;94(3):627-638.e1. doi: 10.1016/j.gie.2021.03.936. Epub 2021 Apr 11.
Endoscopic submucosal dissection (ESD) and EMR are applied in treating superficial colorectal neoplasms but are contraindicated by deeply invasive colorectal cancer (CRC). The invasion depth of neoplasms can be examined by an automated artificial intelligence (AI) system to determine the applicability of ESD and EMR.
A deep convolutional neural network with a tumor localization branch to guide invasion depth classification was constructed on the GoogLeNet architecture. The model was trained using 7734 nonmagnified white-light colonoscopy (WLC) images supplemented by image augmentation from 657 lesions labeled with histopathologic analysis of invasion depth. An independent testing dataset consisting of 1634 WLC images from 156 lesions was used to validate the model.
For predicting noninvasive and superficially invasive neoplasms, the model achieved an overall accuracy of 91.1% (95% confidence interval [CI], 89.6%-92.4%), with 91.2% sensitivity (95% CI, 88.8%-93.3%) and 91.0% specificity (95% CI, 89.0%-92.7%) at an optimal cutoff of .41 and the area under the receiver operating characteristic (AUROC) curve of .970 (95% CI, .962-.978). Inclusion of the advanced CRC data significantly increased the sensitivity in differentiating superficial neoplasms from deeply invasive early CRC to 65.3% (95% CI, 61.9%-68.8%) with an AUROC curve of .729 (95% CI, .699-.759), similar to experienced endoscopists (.691; 95% CI, .624-.758).
We have developed an AI-enhanced attention-guided WLC system that differentiates noninvasive or superficially submucosal invasive neoplasms from deeply invasive CRC with high accuracy, sensitivity, and specificity.
内镜黏膜下剥离术（ESD）和内镜黏膜切除术（EMR）适用于治疗表浅性结直肠肿瘤，但对于深层浸润性结直肠癌（CRC）则不适用。肿瘤的浸润深度可以通过自动人工智能（AI）系统进行检查，以确定 ESD 和 EMR 的适用性。
在 GoogLeNet 架构上构建了一个带有肿瘤定位分支的深度卷积神经网络，用于指导浸润深度分类。该模型使用 7734 张非放大白光结肠镜（WLC）图像进行训练，并通过对 657 个病变进行图像增强来补充，这些病变的浸润深度均经过组织病理学分析标记。使用包含 156 个病变的 1634 张 WLC 图像的独立测试数据集来验证模型。
对于预测非浸润性和表浅浸润性肿瘤，该模型的总体准确率为 91.1%（95%置信区间[CI]，89.6%-92.4%），在最佳截断值为.41 时，灵敏度为 91.2%（95%CI，88.8%-93.3%），特异性为 91.0%（95%CI，89.0%-92.7%），ROC 曲线下面积（AUROC）为.970（95%CI，.962-.978）。纳入高级 CRC 数据可显著提高区分表浅肿瘤与深层浸润性早期 CRC 的灵敏度，达到 65.3%（95%CI，61.9%-68.8%），AUROC 曲线为.729（95%CI，.699-.759），与经验丰富的内镜医生相当（.691；95%CI，.624-.758）。
我们开发了一种 AI 增强注意力引导的 WLC 系统，该系统可以准确、敏感、特异性地区分非浸润性或表浅黏膜下浸润性肿瘤与深层浸润性 CRC。
