文献速递胸部技术与安全CT

光谱CT瘤内及瘤周影像组学预测肺腺癌磨玻璃结节浸润性

Spectral CT-based intratumoral and peritumoral radiomics for predicting invasiveness of ground-glass nodules in lung adenocarcinoma.
PubMed 文献 · Radiol Med · 2026-06-15

结果显示联合瘤内与瘤周多参数模型对AIS、MIA、IAC的内部AUC分别为0.814、0.863、0.988,外部验证AUC分别为0.898、0.774、0.812。

要点

  • 联合瘤内与2mm瘤周多参数影像组学可提高AIS、MIA与IAC的无创鉴别效能
  • 光谱CT衍生的有效原子序数和电子密度图可提供额外诊断价值
  • 模型对IAC的鉴别效能最高,外部验证AUC达0.812

核心结论

  • 联合瘤内与瘤周多参数光谱CT影像组学可提高肺腺癌磨玻璃结节AIS、MIA、IAC的无创鉴别效能
  • 2mm瘤周多参数融合模型在所有瘤周模型中表现最佳,内部宏平均AUC为0.795,外部验证宏平均AUC为0.768

研究设计与样本

  • 回顾性多中心研究,中心1纳入241例GGN患者(AIS=42,MIA=138,IAC=61),中心2作为外部验证纳入87例(AIS=35,MIA=27,IAC=25)
  • 中心1按7:3随机分为训练集和测试集
  • 所有患者均接受双层光谱探测器CT平扫
  • 影像组学特征提取自瘤内及1-5mm瘤周区域,基于常规多色图像、有效原子序数图及电子密度图

诊断标准与关键数值

  • 2mm瘤周多参数融合模型内部宏平均AUC为0.795,外部验证宏平均AUC为0.768
  • 联合瘤内与瘤周多参数模型对AIS内部AUC为0.814,外部AUC为0.898
  • 联合瘤内与瘤周多参数模型对MIA内部AUC为0.863,外部AUC为0.774
  • 联合瘤内与瘤周多参数模型对IAC内部AUC为0.988,外部AUC为0.812

临床应用与阅片要点

  • 平扫光谱CT的常规多色图像、有效原子序数图及电子密度图均可用作影像组学特征来源
  • 结合瘤周2mm区域的影像组学特征可提升对磨玻璃结节浸润性的预测效能
  • 该模型可作为术前无创判断肺腺癌磨玻璃结节病理亚型的决策支持工具

局限性

  • 本研究为回顾性设计
  • 样本中AIS和IAC数量相对MIA较少

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: To evaluate the potential of spectral detector computed tomography (SDCT) combined with intratumoral and peritumoral radiomics for noninvasive characterization of lung adenocarcinomas (LUAD) presenting as ground-glass nodules (GGNs).

METHODS: This retrospective multicenter study included patients from two centers. In Center 1, 241 patients with GGNs (adenocarcinoma in situ [AIS], n = 42; minimally invasive adenocarcinoma [MIA], n = 138; invasive adenocarcinoma [IAC], n = 61) were randomly split into training and test sets (7:3). Center 2 served as an external validation cohort including 87 patients (AIS, n = 35; MIA, n = 27; IAC, n = 25). All patients underwent unenhanced dual-layer SDCT examinations. Radiomics features were extracted from intratumoral and peritumoral regions (1-5 mm) based on SDCT-derived conventional polychromatic images (CPI), effective atomic number (Zeff), and electron density (ED) maps. Machine learning models were then developed and evaluated. Model performance was assessed using the area under the receiver operating characteristic curve (AUC).

RESULTS: Among all peritumoral models, the 2 mm peritumoral multiparameter fusion model showed the best performance, with macro-average AUCs of 0.795 and 0.768 in the internal and external validation sets, respectively. Combining intratumoral and peritumoral multiparameter models further improved performance, achieving internal AUCs of 0.814, 0.863, and 0.988 and external AUCs of 0.898, 0.774, and 0.812 for AIS, MIA, and IAC, respectively.

CONCLUSIONS: The combination of intratumoral and peritumoral multiparametric radiomics derived from SDCT may enhance the noninvasive differentiation of LUAD presenting as GGNs, potentially serving as a valuable tool for supporting clinical decision-making.

来源:Radiol Med,查看原文。版权归原作者所有。

引用

  1. Yang, Kang, Zhao, et al. Spectral CT-based intratumoral and peritumoral radiomics for predicting invasiveness of ground-glass nodules in lung adenocarcinoma. Radiol Med. 2026. PMID: 42295586. doi:10.1007/s11547-026-02239-4

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。