文献速递消化内镜消化道肿瘤与早癌炎症性肠病

前瞻性内镜视频数据库用于AI模型开发、验证和临床结局研究

Implementation of a prospective video databank for artificial intelligence model development, validation, and clinical outcomes research in gastrointestinal endoscopy.
PubMed 文献 · Gastrointest Endosc · 2026-06-25

本研究自2022年起在一家三级中心前瞻性连续入组接受胃镜或结肠镜检查的患者,建立标准化标注的内镜视频数据库。截至2026年3月,共纳入8658例患者、10831例次操作,每例结肠镜和胃镜分别最多记录747和455个结构化变量,病理结果存在于83.7%的记录中。该数据库整合全长视频、实时操作标注和组织病理学关联,为临床结局研究和AI开发验证提供可溯源的数据基础设施。

要点

  • 前瞻性连续入组接受胃镜或结肠镜检查的患者,采集全长去标识化视频并实时结构化标注。
  • 截至2026年3月纳入8658例患者、10831例次操作,由55名内镜医师完成。
  • 每个息肉和活检分别标注49和45个变量,炎症性肠病专项每例操作最多188个变量。

核心结论

  • 该前瞻性数据库整合全长视频、实时操作标注和组织病理学关联,为需要可溯源数据的内镜临床结局研究和AI开发验证提供基础设施。

研究设计与样本

  • 自2022年起在一家三级中心前瞻性连续入组接受胃镜或结肠镜检查的患者(NCT06822616)。
  • 内镜操作以1920×1080分辨率、60帧/秒、10位色深全长录制,采集时去标识化。
  • 培训人员在操作过程中于结构化病例报告表中记录带时间戳的解剖标志、病变特征、光学诊断、干预和疾病评分。
  • 病理结果和元数据与内镜观察事件及对应时间戳视频帧关联。

主要发现

  • 截至2026年3月,共纳入8658例患者(平均年龄59.3岁,女性52.0%),贡献10831例次操作(结肠镜7909例次,胃镜2922例次),由55名内镜医师完成。
  • 每例结肠镜和胃镜分别最多采集747和455个结构化变量。
  • 单个息肉(n=10057)和活检(n=11346)分别标注49和45个变量。
  • 炎症性肠病专项记录每例操作最多包含188个变量。
  • 83.7%的记录中存在病理发现。

临床怎么用

  • 可用于需要可溯源数据的内镜临床结局研究。
  • 可用于AI模型的开发与验证,支持监管合规的AI验证。

局限性

  • 原文未报告局限性。

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: High-quality, annotated data sets are fundamental to clinical research and artificial intelligence (AI) model development. Existing endoscopic databanks are predominantly retrospective, image-based, or lacking patient consent and structured prospective clinical annotations, limiting scientific reproducibility and regulatory-compliant AI validation. This study establishes a prospective endoscopy video databank with standardized annotations enabling AI development and clinical outcomes research. METHODS: Since 2022, we have prospectively enrolled consecutive patients undergoing EGD or colonoscopy at a tertiary center (NCT06822616). Endoscopies were recorded in full (1920 × 1080 resolution, 60 frames per second, 10-bit color depth) and deidentified at acquisition. Trained staff documented timestamped anatomical landmarks, lesion characteristics, optical diagnoses, interventions, and disease scores in structured case report forms during procedures. Pathology results and metadata were linked to all events observed during the endoscopies and corresponding timestamped video frames. RESULTS: Up to March 2026, a total of 8658 patients (mean age, 59.3 years; 52.0% female) were enrolled, contributing 10,831 procedures (7909 colonoscopies and 2922 EGDs) performed by 55 endoscopists. Each colonoscopy and EGD captured up to 747 and 455 structured variables, respectively. Individual polyps (n = 10,057) and biopsies (n = 11,346) were annotated with 49 and 45 variables, respectively. Inflammatory bowel disease-specific documentation comprised up to 188 variables per procedure. Pathologic findings were present in 83.7% of records. CONCLUSIONS: This prospective databank provides the infrastructure for endoscopy research requiring traceable data sourcing, whether for clinical outcomes research or AI development and validation. The integration of full-length videos, real-time procedural annotation, and histopathologic correlation addresses critical gaps in optimizing clinical outcomes and AI research.

来源:Gastrointest Endosc,查看原文。版权归原作者所有。

引用

  1. Oleksiw M, Djinbachian R, Gefflot C, et al. Implementation of a prospective video databank for artificial intelligence model development, validation, and clinical outcomes research in gastrointestinal endoscopy. Gastrointest Endosc. 2026. PMID: 42349759. doi:10.1016/j.gie.2026.06.036

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。