文献速递肿瘤 PET/CT放射性药物与辐射防护

LLM编排AI智能体实现PET/CT端到端解读与量化的真实世界初步研究

End-to-End PET/CT Interpretation and Quantification with an LLM-Orchestrated AI Agent: A Real-World Pilot Study.
PubMed 文献 · J Nucl Med · 2026-10-01

该研究开发了一个由大语言模型编排的多工具智能体,在170例肺癌分期基线[18F]FDG PET/CT中实现从原始DICOM到结构化报告草稿的全自动流程。原发肿瘤检测灵敏度达100%,但淋巴结和远处转移评估灵敏度与特异性有限,提示需专家监督协作。

要点

  • LLM编排智能体在全部170例中无需人工干预完成端到端工作流。
  • 原发肿瘤检测灵敏度100%,淋巴结灵敏度84.8%、特异性39.4%。
  • 远处转移检测灵敏度70.2%、特异性65.0%,存在系统性误判。

核心结论

  • LLM编排的PET/CT智能体可在真实世界条件下实现从原始DICOM到量化及结构化报告草稿的工作流级自动化。
  • 原发肿瘤检测高度可靠,但淋巴结和转移评估存在系统性局限,支持在复杂临床场景中与专家监督协作的角色。

研究设计与样本

  • 概念验证研究,开发基于推理的文本LLM编排多工具智能体,完成系列选择、配准与SUV转换、分割检测、MIP生成、视觉LLM解读及结构化报告草稿合成。
  • 回顾性纳入170例因肺癌分期接受基线[18F]FDG PET/CT的患者,以临床报告为参考标准进行评估。

主要发现

  • 智能体在全部170例检查中无人工干预成功完成从原始DICOM选择到结构化报告草稿生成的全流程。
  • 原发肿瘤检测灵敏度为100%。
  • 淋巴结受累检测灵敏度84.8%、特异性39.4%;远处转移检测灵敏度70.2%、特异性65.0%。
  • 对58例淋巴结和57例转移不匹配病例的分析显示,假阳性与反应性或生理性摄取相关,假阴性涉及小体积或解剖不典型转移。

临床怎么用

  • 可将该智能体用于PET/CT工作流自动化及结构化报告草稿生成,但淋巴结和转移评估结果需专家复核。
  • 在复杂临床场景中应保持专家监督下的协作模式,不宜完全依赖智能体独立判断。

局限性

  • 为回顾性概念验证研究,样本仅限肺癌分期基线[18F]FDG PET/CT患者。
  • 淋巴结和远处转移评估的灵敏度与特异性有限,存在系统性假阳性和假阴性。

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

Although deep learning models have improved individual PET analysis, image processing, and quantification tasks, end-to-end automation from raw DICOM data to quantitative clinical reporting remains limited, particularly in heterogeneous real-world settings. Methods: As a proof-of-concept, an autonomous large language model (LLM)-orchestrated multitool agent for end-to-end PET/CT interpretation was developed. A reasoning-based, text-based LLM selected appropriate series from raw DICOM, coordinated registration and SUV conversion, invoked segmentation and detection tools, generated maximum-intensity projections, called a vision-enabled LLM for interpretation, and synthesized structured draft reports. The system was retrospectively evaluated in 170 patients undergoing baseline [18F]FDG PET/CT for lung cancer staging, with clinical reports serving as the reference standard. Results: The agent successfully completed the full end-to-end workflow from raw DICOM selection to structured draft report generation without human intervention in all 170 examinations. Primary tumor detection achieved 100% sensitivity. For nodal involvement, sensitivity was 84.8% and specificity was 39.4%, whereas distant metastasis detection showed 70.2% sensitivity and 65.0% specificity. Discrepancy analysis of 58 nodal and 57 metastatic mismatch cases revealed systematic false-positive findings related to reactive or physiologic uptake and false-negative findings involving small-volume or anatomically atypical metastases. Conclusion: LLM-orchestrated PET/CT agents enabled workflow-level automation from raw DICOM to quantification and structured draft reporting under real-world conditions. Although primary tumor detection was highly reliable, nodal and metastatic assessment revealed systematic limitations, supporting a collaborative role with continued expert oversight in complex clinical scenarios.

来源:J Nucl Med,查看原文。版权归原作者所有。

引用

  1. Choi H, Bae S, Na KJ, et al. End-to-End PET/CT Interpretation and Quantification with an LLM-Orchestrated AI Agent: A Real-World Pilot Study. J Nucl Med. 2026. PMID: 42315314. doi:10.2967/jnumed.126.272362

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。