文献速递技术与安全

基础模型在美国放射学中的兴起:关于临床效用、安全性与评估的叙述性综述

The Emergence of Foundation Models in U.S. Radiology: A Narrative Review of Clinical Utility, Safety, and Evaluation.
PubMed 文献 · Acad Radiol · 2025-11-19

该叙述性综述综合2022年1月至2025年5月PubMed、Scopus、arXiv和IEEE Xplore的文献,探讨基础模型在美国放射学中的发展、效用与实施挑战。CheXzero、BioMedCLIP和Med-PaLM等美国模型表现出较强的诊断和报告性能,但面临缺乏FDA许可、外部验证有限及与PACS/RIS系统整合障碍等局限。安全方面持续存在幻觉、自动化偏差和边缘案例表现不佳等问题,作者提出包含持续监测、公平导向设计和国家基础模型注册的实施路线图。

要点

  • 美国基础模型如CheXzero、BioMedCLIP和Med-PaLM诊断与报告性能较强,但均无FDA许可。
  • 安全风险包括幻觉、自动化偏差及边缘案例表现不佳,需持续监测。
  • 与PACS/RIS整合障碍和外部验证有限制约临床部署。

核心结论

  • 基础模型在放射学中展现多模态多任务潜力,但临床就绪度、安全性和监管监督仍存疑问,需持续监测、公平导向设计和国家注册来指导负责任部署。

研究设计与样本

  • 叙述性综述,检索PubMed、Scopus、arXiv和IEEE Xplore 2022年1月至2025年5月的文献,综合架构趋势、临床应用、评估方法和监管进展。

诊断标准与关键数值

  • CheXzero、BioMedCLIP和Med-PaLM等美国模型表现出较强的诊断和报告性能。
  • 上述模型缺乏FDA许可,外部验证有限,并与PACS/RIS系统存在整合障碍。

临床应用与阅片要点

  • 作者提出路线图包括持续监测、公平导向设计和国家基础模型注册以指导负责任部署。
  • 人类在环框架、联邦学习和新兴报告标准显示出前景,但机构准备度和监管清晰度仍不完整。

局限性

  • 美国模型缺乏FDA许可,外部验证有限,并与PACS/RIS系统存在整合障碍。
  • 安全方面持续存在幻觉、自动化偏差和边缘案例表现不佳等问题。
  • 机构准备度和监管清晰度仍不完整。

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)。

英文原文(摘要)· 点击展开

Foundation Models (FMs) mark a significant evolution in medical AI, enabling multimodal and multitask performance across text and imaging. Radiology, with its structured data formats and early adoption of AI, is uniquely positioned to benefit from FM capabilities. However, despite promising technical advances, questions remain about their clinical readiness, safety, and regulatory oversight. This narrative review explores the development, utility, and implementation challenges of FMs in U.S. radiology. Literature from PubMed, Scopus, arXiv, and IEEE Xplore (January 2022 to May 2025) was synthesized to highlight architectural trends, clinical applications, evaluation methods, and regulatory developments. U.S.-based models like CheXzero, BioMedCLIP, and Med-PaLM demonstrate strong diagnostic and reporting performance but face key limitations-including lack of FDA clearance, limited external validation, and integration barriers with PACS/RIS systems. Safety issues such as hallucination, automation bias, and underperformance in edge cases persist. While human-in-the-loop frameworks, federated learning, and emerging reporting standards show promise, institutional readiness and regulatory clarity remain fragmented. We propose a roadmap that includes continuous monitoring, equity-focused design, and a national FM registry to guide responsible deployment. Radiology's digital maturity makes it a critical testbed for foundational AI integration-offering lessons for broader clinical adoption.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Uppuluri PC, Yang CW. The Emergence of Foundation Models in U.S. Radiology: A Narrative Review of Clinical Utility, Safety, and Evaluation. Acad Radiol. 2025. PMID: 41266222. doi:10.1016/j.acra.2025.10.063

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。