与卷积神经网络诊断和分类相关的亚实性肺结节人类可识别CT图像特征
该研究回顾性收集直径≤3 cm的亚实性结节CT图像,用5折交叉验证训练卷积神经网络将其分为良性及浸润前病变、微浸润腺癌和浸润性腺癌三类,准确率93%。通过Grad-CAM和DeepDream可视化发现,良性及浸润前病变的激活区域主要为光滑边缘和磨玻璃成分,浸润性腺癌主要为部分实性和实性成分、分叶及空气支气管征,而微浸润腺癌的激活区域多变。
要点
- CNN分类亚实性结节的准确率达93%,Grad-CAM定位的激活区域与放射科医生诊断经验一致。
- 良性及浸润前病变的CNN激活特征以光滑边缘和磨玻璃成分为主。
- 浸润性腺癌的CNN激活特征以部分实性/实性成分、分叶和空气支气管征为主。
核心结论
- CNN在CT图像上对亚实性结节三分类(良性及浸润前病变、微浸润腺癌、浸润性腺癌)的准确率为93%。
- Grad-CAM定位的激活区域与放射科医生诊断亚实性结节的专业经验一致,支持人工智能的临床应用。
研究设计与样本
- 回顾性研究,共纳入569例患者的586个直径≤3 cm的亚实性结节,采用5折交叉验证训练和验证CNN。
- 参考标准为组织学证实或随访6.4年,结节分为良性及浸润前病变346个、微浸润腺癌144个、浸润性腺癌96个。
- 使用梯度加权类激活图(Grad-CAM)、分离的激活通道和区域以及DeepDream算法研究CNN机制。
关键影像表现/征象
- 良性及浸润前病变组的激活区域主要为光滑边缘(p < 0.001)和磨玻璃成分(p = 0.033)。
- 浸润性腺癌组的激活区域主要为部分实性(p < 0.001)和实性成分(p < 0.001)、分叶状(p < 0.001)及空气支气管征(p < 0.001)。
- 微浸润腺癌的激活区域多变。
- DeepDream算法以人类可识别的方式展示了CNN从训练数据集中学到的图像特征。
诊断标准与关键数值
- CNN将亚实性结节分为良性及浸润前病变、微浸润腺癌和浸润性腺癌三类的准确率为93%。
- Grad-CAM成功定位决定最终分类的整个图像特征区域。
- 良性及浸润前病变的激活区域与光滑边缘和磨玻璃成分相关,浸润性腺癌的激活区域与部分实性/实性成分、分叶和空气支气管征相关。
临床应用与阅片要点
- 阅片时可关注光滑边缘和磨玻璃成分,这些是CNN判为良性及浸润前病变的重要激活特征。
- 部分实性/实性成分、分叶和空气支气管征是CNN判为浸润性腺癌的主要激活特征,报告时应注意描述。
- 微浸润腺癌的CNN激活区域多变,影像表现不典型时需结合多征象综合判断。
局限性
- 回顾性研究设计。
- 单中心数据,样本量有限(569例患者,586个结节)。
- 微浸润腺癌的激活区域多变,CNN在该类别的可解释性仍不足。
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
OBJECTIVES: The interpretability of convolutional neural networks (CNNs) for classifying subsolid nodules (SSNs) is insufficient for clinicians. Our purpose was to develop CNN models to classify SSNs on CT images and to investigate image features associated with the CNN classification.
METHODS: CT images containing SSNs with a diameter of ≤ 3 cm were retrospectively collected. We trained and validated CNNs by a 5-fold cross-validation method for classifying SSNs into three categories (benign and preinvasive lesions [PL], minimally invasive adenocarcinoma [MIA], and invasive adenocarcinoma [IA]) that were histologically confirmed or followed up for 6.4 years. The mechanism of CNNs on human-recognizable CT image features was investigated and visualized by gradient-weighted class activation map (Grad-CAM), separated activation channels and areas, and DeepDream algorithm.
RESULTS: The accuracy was 93% for classifying 586 SSNs from 569 patients into three categories (346 benign and PL, 144 MIA, and 96 IA in 5-fold cross-validation). The Grad-CAM successfully located the entire region of image features that determined the final classification. Activated areas in the benign and PL group were primarily smooth margins (p < 0.001) and ground-glass components (p = 0.033), whereas in the IA group, the activated areas were mainly part-solid (p < 0.001) and solid components (p < 0.001), lobulated shapes (p < 0.001), and air bronchograms (p < 0.001). However, the activated areas for MIA were variable. The DeepDream algorithm showed the image features in a human-recognizable pattern that the CNN learned from a training dataset.
CONCLUSION: This study provides medical evidence to interpret the mechanism of CNNs that helps support the clinical application of artificial intelligence.
KEY POINTS: • CNN achieved high accuracy (93%) in classifying subsolid nodules on CT images into three categories: benign and preinvasive lesions, MIA, and IA. • The gradient-weighted class activation map (Grad-CAM) located the entire region of image features that determined the final classification, and the visualization of the separated activated areas was consistent with radiologists' expertise for diagnosing subsolid nodules. • DeepDream showed the image features that CNN learned from a training dataset in a human-recognizable pattern.
引用
- Jiang B, Zhang Y, Zhang L, et al. Human-recognizable CT image features of subsolid lung nodules associated with diagnosis and classification by convolutional neural networks. Eur Radiol. 2021. PMID: 33847813. doi:10.1007/s00330-021-07901-1
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。