放射学重塑:影像AI实践演示中的互操作性与经验
本文介绍RSNA年会「Radiology Reimagined」演示项目,展示如何利用互操作性和语义标准将人工智能整合入放射工作流,涵盖从检查申请到计费的全流程。自2020年起,演示场景从单纯肺结节和颅内出血检测,发展到2024年前列腺MRI判读和淀粉样相关影像异常检测。该项目强调DICOM、HL7 FHIR、IHE框架等标准在促进AI临床应用中的作用。
要点
- AI整合需覆盖检查申请、扫描、诊断、报告、随访全工作流。
- 关注DICOM、HL7 FHIR、FHIRcast等标准对AI落地的意义。
- 从单一病灶检测向器官特异性综合判读发展是AI应用趋势。
核心结论
- 互操作性和语义标准是推动AI进入放射科实际工作流的关键。
研究设计与样本
- 基于RSNA年会的持续性技术演示项目,自2020年起每年展示。
- 由放射科与超过20家行业厂商协作,采用真实临床场景进行演示。
关键影像表现/征象
- 2020年演示场景为单纯肺结节检测和颅内出血检测。
- 2024年演示场景发展为前列腺MRI判读和淀粉样相关影像异常检测。
诊断标准与关键数值
- 采用的标准包括DICOM、HL7 FHIR、IHE框架、Integrated Multimedia Reporting及FHIRcast。
临床应用与阅片要点
- AI项目实施需重视与现有PACS/RIS及临床信息系统的双向集成。
- 报告环节可引入集成多媒体报告标准,提升结构化与互操作能力。
局限性
- 演示场景并非真实临床部署,定量效果反馈仍在完善中。
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
The Radiology Reimagined demonstration, formerly known as Imaging AI in Practice, is presented at the Radiological Society of North America (RSNA) Annual Meeting. It highlights how interoperability and semantic standards can facilitate the integration of artificial intelligence (AI) into radiology workflows. This integration encompasses examination ordering, scheduling, protocoling, image acquisition, examination interpretation, reporting, results communication, follow-up, and billing. Since 2020, Radiology Reimagined has become a prominent feature of the RSNA Annual Meeting, showcasing AI's role in clinical imaging workflows. The demonstration fosters collaboration between radiology and industry vendor partners, promoting the adoption of standards such as Digital Imaging and Communications in Medicine, Health Level 7 Fast Healthcare Interoperability Resources (FHIR), and Integrating the Healthcare Enterprise profiles. Realistic clinical scenarios demonstrate the utility and challenges of integrating AI tools into point-of-care systems. The complexity of Radiology Reimagined scenarios has increased, from simple pulmonary nodule and intracranial hemorrhage detection in 2020 to prostate MRI interpretation and amyloid-related imaging abnormality detection in 2024. Each year, more than 20 unique vendor partners participate, progressively incorporating new standards such as Integrated Multimedia Reporting in 2023 and improved support for FHIRcast (which synchronizes health care applications in real time) in 2024. Participants learn how to facilitate interoperability and integrate software into their clinical workflows, with an emphasis on practical advice and key considerations. Future iterations of the demonstration will aim to expand attendee engagement, formalize quantitative feedback, and broaden the use of emerging standards. Videos of the annual demonstrations are available on the RSNA website.
引用
- Jiao, Hussain, Sippel Schmidt, et al. Radiology Reimagined: Interoperability and Lessons Learned from the Imaging AI in Practice Demonstration. Radiology. 2026. PMID: 42301015. doi:10.1148/radiol.252225
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。