文献速递呼吸道感染肺结节与肺癌

胸部影像人工智能在新冠肺炎诊断中的应用与未来研究方向

[Chest imaging-based artificial intelligence in the diagnosis of coronavirus disease 2019 and prospects for future research].
PubMed 文献 · Zhonghua Jie He He Hu Xi Za Zhi · 2022-01-01

本文综述胸部影像人工智能在肺炎诊断中的研究进展,重点介绍其在新冠肺炎快速识别、病灶定量分析及预后评估中的临床价值,并指出现有研究存在样本量小、模型缺乏多模式评估、肺炎分类欠精细等不足,提出未来研究方向。

要点

  • 胸部影像AI可辅助新冠肺炎的快速识别与病灶定量分析。
  • AI系统有助于识别需密切监测和及时干预的高风险新冠肺炎患者。
  • 现有研究存在样本量小、缺乏多模式评估、肺炎分类粗糙等局限。

核心结论

  • 胸部影像AI有望在未来肺炎临床决策中发挥重要作用。

研究设计与样本

  • 本文为综述,总结胸部影像AI在肺炎(包括新冠肺炎)诊断中的研究现状并提出未来研究建议。

主要发现

  • 胸部影像AI在新冠肺炎快速识别、病灶定量分析、疾病严重程度及预后评估方面展示出临床价值。
  • AI系统可辅助临床医生识别需要密切监测和及时干预的高风险新冠肺炎患者。
  • 现有研究存在样本量小、AI模型缺乏多模式评估、肺炎分类欠精细等局限性。

关键数值与推荐

  • 未来研究应重视高质量数据集的采集、影像数据标注的标准化、技术创新、算法优化和AI模型验证。
  • 建议加强影像AI在病毒性肺炎、细菌性肺炎和真菌性肺炎等其他类型肺炎中的研究。

临床怎么用

  • 临床可关注胸部影像AI在新冠肺炎快速识别、病灶定量及预后评估中的辅助价值。
  • 应用AI系统辅助识别需密切监测和及时干预的高风险新冠肺炎患者。

局限性

  • 现有研究样本量小。
  • AI模型缺乏多模式评估。
  • 肺炎分类欠精细。

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

Artificial intelligence (AI) has been applied increasingly in the medical field during the past 5 years. Within respiratory medicine, chest imaging AI is one of the relevant hotspots, commonly trained to identify pulmonary nodules/lung tumors, tuberculosis, pneumonia, interstitial lung disease, chronic obstructive pulmonary disease, pulmonary embolism and other pathologies. Due to the non-specific clinical manifestations and the low detection rate of pathogens, precise diagnosis and treatment of pneumonia remain challengeable. Since the outbreak of coronavirus disease 2019 (COVID-19), chest imaging AI has demonstrated its clinical value in accurate diagnosis and quantitative measurements of COVID-19. Moreover, an AI system can assist the clinicians to identify the high-risk COVID-19 patients who warrant close monitoring and timely intervention. However, there are still some limitations in the existing studies, such as small sample size, lack of multi-modal assessment of the AI model, and rough classification of pneumonia. Therefore, some suggestions for future research were put forward in this paper. Most of all, more attention should be paid to the collection of high-quality datasets, standardization of image annotation, technology innovation, algorithm optimization and model verification. Besides, the application of imaging AI on other types of pneumonia including viral pneumonia, bacterial pneumonia and pneumomycosis deserves further study. In conclusion, chest imaging AI is expected to play a vital role in decision-making for pneumonia in the future. 人工智能(AI)在医学领域的应用研究越来越多,影像AI是最受关注的热点之一。鉴于临床表现缺乏特异性、病原检测率低等因素,肺炎的精准诊疗面临巨大挑战。新型冠状病毒肺炎(简称新冠肺炎)疫情暴发以来,胸部影像AI展示了其在新冠肺炎快速识别、病灶定量分析、疾病严重程度及预后评估等方面的价值,但仍存在一些不足,如研究样本量小,模型缺乏多模式评估,肺炎分类欠精细等。本文在此基础上,对影像AI辅助肺炎诊断的今后研究提出一些建议,强调高质量数据集的采集、影像数据标注的标准化、技术创新、算法优化和AI模型的验证,以及重视AI在其他类型肺炎中的研究,期待影像AI为肺炎的临床决策提供更多参考。.

来源:Zhonghua Jie He He Hu Xi Za Zhi,查看原文。版权归原作者所有。

引用

  1. Li Y, Liu SY, Zheng JP, et al. [Chest imaging-based artificial intelligence in the diagnosis of coronavirus disease 2019 and prospects for future research]. Zhonghua Jie He He Hu Xi Za Zhi. 2022. PMID: 36480857. doi:10.3760/cma.j.cn112147-20220519-00429

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。