文献速递肝病消化道肿瘤与早癌消化内镜

BCLC分期与AI图像定量:必将融合,但如何与何时?

BCLC classification and AI-based image quantification: What is meant to be will come together - but how and when?
PubMed 文献 · J Hepatol · 2026-03-06

本文评述肝细胞癌BCLC分期体系与人工智能图像定量方法的协同潜力。BCLC基于简单临床与影像参数,未充分利用常规影像信息;AI可高精度自动提取定量影像特征,但技术、管理、成本、培训及标准化验证与工作流整合等障碍导致转化差距持续存在。

要点

  • BCLC仍是肝细胞癌预后评估与初始治疗选择的主要体系,获EASL指南重申。
  • AI可自动定量影像特征,有望与BCLC互补用于临床决策。
  • AI在肿瘤体积测量、影像组学等方面进展显著,但临床转化仍存差距。

核心结论

  • AI定量影像有潜力与BCLC分类协同,但标准化验证缺失与工作流整合不足阻碍其常规临床落地。

研究设计与样本

  • 本文为综述性文章,评估当前AI定量影像参数及其与BCLC分类协同的潜力。

主要发现

  • BCLC分类二十余年来是肝细胞癌预后评估与初始治疗选择的主要体系,并被最新EASL肝细胞癌管理临床实践指南重申。
  • BCLC设计基于简单临床与影像参数,临床适用性高,但未充分利用常规放射影像中潜在编码的全部信息。
  • AI方法已成熟,可在较少用户输入下高精度全自动提取和定量数字影像特征。
  • AI在自动肿瘤体积测量、影像组学、转移灶检测及机会性影像生物标志物识别等领域取得实质进展,但仍存在转化差距。
  • 技术、管理、成本与培训因素,以及抵触变革、缺乏标准化AI验证和工作流整合缺失,均妨碍常规临床实施。

临床怎么用

  • 在BCLC分类框架基础上,可关注AI定量影像参数作为补充信息用于肝细胞癌临床决策的潜在价值。

局限性

  • AI临床实施受技术、管理、成本与培训因素制约,并存在抵触变革、缺乏标准化验证及工作流整合缺失等障碍。

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

The Barcelona Clinic Liver Cancer (BCLC) classification has been the mainstay for prognostic assessment and initial treatment selection in hepatocellular carcinoma (HCC) for more than two decades. It is widely clinically accepted and has been reaffirmed in the recently updated European Association for the Study of the Liver (EASL) Clinical Practice Guidelines on the management of HCC. Its design is based on simple clinical and imaging parameters, which makes it highly applicable in clinical practice. However, it does not fully utilise all the information potentially encoded in routine radiology imaging. With artificial intelligence (AI) methods now maturing, we have a robust fully automated way to extract and quantify digital imaging features without much user input and with high precision. Therefore, AI could bridge quantitative imaging into clinical decision-making, together with the existing BCLC classification. However, despite substantial AI advancements in many fields such as automated tumour volumetry, radiomics, detection of metastatic lesions, and the identification of opportunistic imaging biomarkers, a translational gap persists. While challenges related to technical, administrative, cost, and training factors have to be taken into account, a certain aversion to change, as well as an absence of standardised AI validation and missing workflow integration hamper clinical implementation in routine care. This article aims to evaluate current AI-quantified imaging parameters and their potential for synergy with the established BCLC classification.

来源:J Hepatol,查看原文。版权归原作者所有。

引用

  1. Müller L, Kather JN, Marquardt JU, et al. BCLC classification and AI-based image quantification: What is meant to be will come together - but how and when? J Hepatol. 2026. PMID: 41794137. doi:10.1016/j.jhep.2026.02.027

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。