人工智能用于肺部氙气通气MRI的疾病分类
研究利用六种肺部疾病(对照、哮喘、闭塞性细支气管炎综合征、支气管肺发育不良、囊性纤维化、淋巴管肌瘤病)的氙气通气MRI数据训练卷积神经网络,262名参与者的训练/测试数据中,最佳网络VGG-16的top-1准确率为56%、top-2准确率为78%、AUC为0.85,且在单训练折上网络top-1准确率61%优于人类观察者的40%。
要点
- 氙气通气MRI的VDP单独不能区分疾病,但人工智能网络可基于空间模式分类疾病。
- 最佳网络VGG-16的top-1准确率56%、top-2准确率78%、AUC 0.85。
- 网络在较大类别上表现更好:对照62%、囊性纤维化67%、淋巴管肌瘤病69%。
核心结论
- 开发的人工智能工具可仅凭氙气通气图像分类疾病,且优于人类观察者,提示氙气图像含有额外的疾病特异性信息,可能有助于临床疑难病例或疾病表型分析。
研究设计与样本
- 使用六种肺部疾病(对照、哮喘、闭塞性细支气管炎综合征、支气管肺发育不良、囊性纤维化、淋巴管肌瘤病)的氙气MRI数据训练卷积神经网络,训练/测试数据来自262名参与者。
关键影像表现/征象
- 既往研究报道了疾病特异性缺陷模式的轶事证据,如哮喘的楔形缺陷和淋巴管肌瘤病的圆点状缺陷。
- 梯度类激活映射用于可视化图像中对分类重要的部分。
诊断标准与关键数值
- 网络性能评估指标包括top-1准确率、top-2准确率、召回率、精确率和一对多AUC。
- 最佳网络VGG-16的top-1准确率=56%,top-2准确率=78%,召回率=.30,精确率=.70,AUC=.85。
- 网络在较大类别上top-1准确率:对照=62%(n=57),囊性纤维化=67%(n=85),淋巴管肌瘤病=69%(n=61)。
- 单训练折上人类观察者top-1准确率=40%,网络top-1准确率=61%。
临床应用与阅片要点
- 氙气通气MRI图像可用于人工智能分类疾病,网络表现优于人类观察者。
- 梯度类激活映射可帮助理解网络分类依据的图像区域。
局限性
- 原文未报告局限性。
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)。
英文原文(摘要)· 点击展开
RATIONALE AND OBJECTIVES: Hyperpolarized 129Xenon magnetic resonance imaging (MRI) measures the extent of lung ventilation by ventilation defect percent (VDP), but VDP alone cannot distinguish between diseases. Prior studies have reported anecdotal evidence of disease-specific defect patterns such as wedge-shaped defects in asthma and polka-dot defects in lymphangioleiomyomatosis (LAM). Neural network artificial intelligence can evaluate image shapes and textures to classify images, but this has not been attempted in xenon MRI. We hypothesized that an artificial intelligence network trained on ventilation MRI could classify diseases based on spatial patterns in lung MR images alone.
MATERIALS AND METHODS: Xenon MRI data in six pulmonary conditions (control, asthma, bronchiolitis obliterans syndrome, bronchopulmonary dysplasia, cystic fibrosis, LAM) were used to train convolutional neural networks. Network performance was assessed with top-1 and top-2 accuracy, recall, precision, and one-versus-all area under the curve (AUC). Gradient class-activation-mapping (Grad-CAM) was used to visualize what parts of the images were important for classification.
RESULTS: Training/testing data were collected from 262 participants. The top performing network (VGG-16) had top-1 accuracy=56%, top-2 accuracy=78%, recall=.30, precision=.70, and AUC=.85. The network performed better on larger classes (top-1 accuracy: control=62% [n=57], CF=67% [n=85], LAM=69% [n=61]) and outperformed human observers (human top-1 accuracy=40%, network top-1 accuracy=61% on a single training fold).
CONCLUSION: We developed an artificial intelligence tool that could classify disease from xenon ventilation images alone that outperformed human observers. This suggests that xenon images have additional, disease-specific information that could be useful for cases that are clinically challenging or for disease phenotyping.
引用
- Matheson AM, Bdaiwi AS, Willmering MM, et al. Disease Classification of Pulmonary Xenon Ventilation MRI Using Artificial Intelligence. Acad Radiol. 2025. PMID: 40617705. doi:10.1016/j.acra.2025.06.024
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。