文献速递食管胃疾病与幽门螺杆菌消化内镜消化道肿瘤与早癌

内镜人工智能预测浅表上消化道癌浸润深度的诊断准确性荟萃分析

Endoscopic Artificial Intelligence for Invasion-Depth Prediction in Superficial Upper Gastrointestinal Cancer: A Diagnostic Test Accuracy Meta-Analysis.
PubMed 文献 · Gastrointest Endosc · 2026-09-18

该诊断准确性荟萃分析纳入26项研究,其中18项(14项胃腺癌、4项食管鳞癌)进入6568例验证/测试患者或病变的主分析池。AI预测黏膜下浸润的汇总敏感度0.77(95%CI 0.71-0.82)、特异度0.88(95%CI 0.85-0.90),外部验证特异度低于内部验证(0.85 vs 0.93)。结论认为AI准确性具临床意义但依赖验证框架,规则入选价值优于排除价值,近期角色为第二读者支持而非独立判定ESD适应证。

要点

  • AI预测黏膜下浸润汇总敏感度0.77、特异度0.88,阳性似然比6.30、阴性似然比0.26。
  • 外部验证特异度(0.85)低于内部验证(0.93),验证框架是唯一与性能显著相关的协变量。
  • AI单独敏感度高于无辅助内镜医师(82.5% vs 63.8%,+18.7个百分点,配对t=3.29,p=0.013)。

核心结论

  • 内镜AI对浅表上消化道癌黏膜下浸润的预测具有临床意义但依赖验证框架的准确性。
  • AI对黏膜下浸润的规则入选(rule-in)价值强于排除(rule-out)价值。
  • 近期最合理的角色是第二读者支持,而非独立判定ESD候选资格。
  • 临床部署前应进行独立患者层面的外部验证。

研究设计与样本

  • 诊断准确性荟萃分析,检索PubMed/MEDLINE、Embase、Cochrane CENTRAL和Web of Science至2026年6月11日。
  • 纳入将AI应用于浅表上消化道癌内镜图像或视频、以组织病理学为标准的研究。
  • 26项研究符合条件,18项(14项胃腺癌、4项食管鳞癌)进入6568例验证/测试患者或病变的主分析池。
  • 采用双变量随机效应模型汇总敏感度和特异度,并进行亚组与meta回归分析。
  • 使用QUADAS-2评估偏倚风险,PROSPERO注册号CRD420261400366。

主要发现

  • 汇总敏感度0.77(95%CI 0.71-0.82),汇总特异度0.88(95%CI 0.85-0.90)。
  • 阳性似然比6.30,阴性似然比0.26。
  • 外部验证特异度低于内部验证(0.85 vs 0.93)。
  • 验证框架是预设meta回归中唯一与性能显著相关的协变量(特异度伪R2 77.9%),AI架构和成像方式不显著。
  • 8项研究内配对比较显示AI单独平均每研究敏感度高于无辅助内镜医师(82.5% vs 63.8%;+18.7个百分点;配对t=3.29,p=0.013),特异度无显著差异;AI辅助读者比较数据有限且为探索性。

关键数值与推荐

  • 汇总敏感度0.77(95%CI 0.71-0.82),特异度0.88(95%CI 0.85-0.90),阳性似然比6.30,阴性似然比0.26。
  • 外部验证特异度0.85 vs 内部验证0.93。
  • 验证框架与特异度相关,伪R2 77.9%。
  • AI单独 vs 无辅助内镜医师敏感度82.5% vs 63.8%,+18.7个百分点,配对t=3.29,p=0.013。

临床怎么用

  • AI可作为第二读者辅助内镜医师预测黏膜下浸润,而非独立决定ESD适应证。
  • AI阳性结果对黏膜下浸润的规则入选价值较强,阴性结果排除价值有限。
  • 临床使用前需关注AI模型是否经过独立患者层面外部验证。

局限性

  • 外部验证特异度低于内部验证,提示性能受验证框架影响。
  • AI辅助读者比较数据有限,第二读者角色仍为探索性。
  • 研究间AI架构和成像方式异质性存在,但meta回归未显示显著关联。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND AND AIMS: Pre-resection prediction of submucosal invasion is pivotal for endoscopic submucosal dissection (ESD) candidate selection in early gastric and esophageal cancers. We performed a diagnostic test accuracy meta-analysis of artificial intelligence (AI) for invasion-depth prediction. METHODS: PubMed/MEDLINE, Embase, Cochrane CENTRAL, and Web of Science were searched through 11 June 2026. Eligible studies applied AI to endoscopic images or videos of superficial upper gastrointestinal cancers with histopathological reference standards. Pooled sensitivity and specificity were estimated with a bivariate random-effects model; subgroup and meta-regression analyses examined validation framework, histology, AI architecture, and imaging modality. Risk of bias was assessed with QUADAS-2. RESULTS: Twenty-six studies were eligible; 18 (14 gastric adenocarcinoma, 4 esophageal squamous cell carcinoma) contributed to the primary pool of 6,568 validation/test patients or lesions. Pooled sensitivity was 0.77 (95% CI 0.71-0.82) and specificity 0.88 (0.85-0.90); positive and negative likelihood ratios were 6.30 and 0.26. Externally validated specificity was lower than internally validated specificity (0.85 vs. 0.93); validation framework was the only covariate significantly associated with performance in prespecified meta-regression (specificity pseudo-R2 77.9%), whereas AI architecture and imaging modality were not. In within-study paired comparisons (n = 8 studies), mean per-study sensitivity was higher for AI alone than for unassisted endoscopists (82.5% vs. 63.8%; +18.7 percentage points; paired t = 3.29, p = 0.013); specificity did not differ significantly. Limited AI-assisted reader-comparison data were directionally consistent with a second-reader role but remained exploratory. CONCLUSIONS: Endoscopic AI shows clinically meaningful, but validation-dependent, accuracy for invasion-depth prediction, supporting stronger rule-in than rule-out utility for submucosal invasion. Its most plausible near-term role is second-reader support rather than stand-alone determination of ESD candidacy; independent patient-level external validation should be considered essential before clinical deployment. (PROSPERO registration: CRD420261400366).

来源:Gastrointest Endosc,查看原文。版权归原作者所有。

引用

  1. Lee H, Min YW, Kim S, et al. Endoscopic Artificial Intelligence for Invasion-Depth Prediction in Superficial Upper Gastrointestinal Cancer: A Diagnostic Test Accuracy Meta-Analysis. Gastrointest Endosc. 2026. PMID: 42759726. doi:10.1016/j.gie.2026.09.022

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。