文献速递消化内镜消化道肿瘤与早癌

人工智能自动评估结肠镜质量指标

Artificial Intelligence-Automated Assessment of Colonoscopy Quality Metrics.
PubMed 文献 · Am J Gastroenterol · 2026-08-01

本研究在单中心回顾性分析了18,597例结肠镜操作,使用人工智能结肠镜质量(AI-CQ)工具分析录像并与护士记录的人工指标比较。结果显示AI计算的退镜时间与人工计算强相关(r=0.91),AI息肉/结肠镜与腺瘤检出率强相关(0.54),并能准确测量息肉切除术式。AI-CQ可基于录像可靠测量常用质量指标。

要点

  • AI-CQ工具可自动计算退镜时间,与人工测量强相关(r=0.91)。
  • AI息肉/结肠镜与腺瘤检出率(r=0.54)及锯齿状病变检出率(r=0.67)强相关。
  • AI-CQ能准确测量冷圈套息肉切除术率(均值84.0%,范围63.0%-95.3%)。

核心结论

  • AI-CQ工具能利用结肠镜录像准确测量常用质量指标,为可靠评估结肠镜质量提供可行新方法。

研究设计与样本

  • 单中心学术医学中心两个内镜中心记录的结肠镜操作,共18,597例由55名主治医师完成。
  • 选取11个月研究期间完成≥100例筛查或监测结肠镜的高 volume 筛查结肠镜医师31名,共12,456例筛查或监测结肠镜。
  • 比较AI-CQ工具从录像计算的指标与护士记录的人工计算指标(腺瘤检出率、退镜时间等)。

主要发现

  • AI计算的插入时间与人工插入时间相关(r=0.60),退镜时间与人工退镜时间强相关(r=0.91)。
  • 结肠镜医师中位腺瘤检出率为43.2%。
  • AI息肉/结肠镜为1.47(标准差±0.54),与腺瘤检出率相关(0.54),与锯齿状病变检出率相关(0.67)。
  • AI-CQ准确测量任何息肉切除术和冷圈套息肉切除术,冷圈套息肉切除术率均值84.0%(范围63.0%-95.3%)。

临床怎么用

  • 可利用AI-CQ工具从常规录像中自动测量退镜时间、息肉/结肠镜等指标,辅助质量监测。
  • AI-CQ可准确评估息肉切除术式使用情况,如冷圈套息肉切除术率。

局限性

  • 研究为单中心学术医学中心,结果外推需谨慎。
  • 部分分析仅限高 volume 筛查结肠镜医师(≥100例筛查或监测结肠镜)。

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

INTRODUCTION: There is a significant intraprovider variability in colonoscopy performance, but this is difficult to measure in routine practice. We describe an artificial intelligence colonoscopy quality (AI-CQ) tool that measures colonoscopy quality through analysis of recorded colonoscopy procedures and compare AI-CQ assessment of quality metrics with manual measurement in a large cohort of colonoscopists. METHODS: Colonoscopy procedures were performed at 1 of 2 endoscopy locations at a single academic medical center. Select analyses were restricted to higher volume screening colonoscopists performing ≥100 screening or surveillance colonoscopies over the 11-month study period. Colonoscopy quality metrics were calculated from recorded colonoscopy videos using the AI-CQ tool and compared with manually calculated metrics (through nurse documentation) including adenoma detection rate (ADR) and withdrawal time (WT). RESULTS: A total of 18,597 colonoscopy procedures performed by 55 unique attendings were recorded with 31 higher volume screening colonoscopists performing 12,456 screening or surveillance colonoscopies (median colonoscopist ADR 43.2%). AI-calculated insertion time and WT strongly correlated with manually calculated insertion time ( r = 0.60) and WT ( r = 0.91), respectively. AI polyps per colonoscopy was 1.47 (SD ± 0.54) and strongly correlated with ADR (0.54) and serrated detection rate (0.67). The AI-CQ accurately measured performance of any polypectomy and cold snare polypectomy with a mean cold snare polypectomy rate of 84.0% (range 63.0%-95.3%). DISCUSSION: The AI-CQ can accurately measure commonly used quality metrics using recorded colonoscopy videos. Use of this AI tool provides a novel feasible approach to reliably measuring colonoscopy quality.

来源:Am J Gastroenterol,查看原文。版权归原作者所有。

引用

  1. Keswani RN, Kaklamanos E, Kristinsdottir K, et al. Artificial Intelligence-Automated Assessment of Colonoscopy Quality Metrics. Am J Gastroenterol. 2026. PMID: 42546167. doi:10.14309/ajg.0000000000004170

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。