Department of Gastroenterology and Hepatology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, Osaka, 565-0871, Japan.
Department of Gastroenterology, Yao Municipal Hospital, Yao, 581-0069, Japan.
J Gastroenterol. 2024 Jul;59(7):543-555. doi: 10.1007/s00535-024-02102-1. Epub 2024 May 7.
We developed an artificial intelligence (AI)-based endoscopic ultrasonography (EUS) system for diagnosing the invasion depth of early gastric cancer (EGC), and we evaluated the performance of this system.
A total of 8280 EUS images from 559 EGC cases were collected from 11 institutions. Within this dataset, 3451 images (285 cases) from one institution were used as a development dataset. The AI model consisted of segmentation and classification steps, followed by the CycleGAN method to bridge differences in EUS images captured by different equipment. AI model performance was evaluated using an internal validation dataset collected from the same institution as the development dataset (1726 images, 135 cases). External validation was conducted using images collected from the other 10 institutions (3103 images, 139 cases).
The area under the curve (AUC) of the AI model in the internal validation dataset was 0.870 (95% CI: 0.796-0.944). Regarding diagnostic performance, the accuracy/sensitivity/specificity values of the AI model, experts (n = 6), and nonexperts (n = 8) were 82.2/63.4/90.4%, 81.9/66.3/88.7%, and 68.3/60.9/71.5%, respectively. The AUC of the AI model in the external validation dataset was 0.815 (95% CI: 0.743-0.886). The accuracy/sensitivity/specificity values of the AI model (74.1/73.1/75.0%) and the real-time diagnoses of experts (75.5/79.1/72.2%) in the external validation dataset were comparable.
Our AI model demonstrated a diagnostic performance equivalent to that of experts.
我们开发了一种基于人工智能（AI）的内镜超声（EUS）系统，用于诊断早期胃癌（EGC）的浸润深度，并对该系统的性能进行了评估。
从 11 家机构共收集了 559 例 EGC 病例的 8280 张 EUS 图像。在该数据集中，来自 1 家机构的 3451 张图像（285 例）被用作开发数据集。AI 模型由分割和分类步骤组成，然后使用 CycleGAN 方法来弥合不同设备采集的 EUS 图像之间的差异。使用与开发数据集来自同一机构的内部验证数据集（1726 张图像，135 例）来评估 AI 模型的性能。外部验证使用来自其他 10 家机构的图像（3103 张图像，139 例）进行。
AI 模型在内部验证数据集的曲线下面积（AUC）为 0.870（95%CI：0.796-0.944）。在诊断性能方面，AI 模型、专家（n=6）和非专家（n=8）的准确性/敏感性/特异性值分别为 82.2%/63.4%/90.4%、81.9%/66.3%/88.7%和 68.3%/60.9%/71.5%。AI 模型在外部验证数据集的 AUC 为 0.815（95%CI：0.743-0.886）。在外部验证数据集中，AI 模型的准确性/敏感性/特异性值（74.1%/73.1%/75.0%）和专家的实时诊断结果（75.5%/79.1%/72.2%）相当。
我们的 AI 模型表现出与专家相当的诊断性能。
