含瘤周组织的影像组学特征与临床标签的联合模型预测T1期非小细胞肺癌淋巴结转移  

The combined model of intratumoral and peritumoral radiomic features and clinical signatures predicting the lymph node metastasis in stage T1 non-small cell lung cancer

作　　者：柴亚婷 李淑 姜星原 吴璠 李叶[1] CHAI Ya-ting;LI Shu;JIANG Xing-yuan;WU Fan;LI Ye(The First Hospital of Jilin University Changchun 130012,China)

机构地区：[1]吉林大学第一医院,吉林长春130012

出　　处：《中国临床医学影像杂志》2021年第7期470-475,共6页Journal of China Clinic Medical Imaging

基　　金：实验室医学影像人工智能重点实验室(项目编号:20200601003JC);吉林省放射医疗科技创新中心(项目编号:20190902016TC)。

摘　　要：目的:探讨含瘤周组织的影像组学特征与临床病理标签的联合模型对预测T1期非小细胞肺癌淋巴结转移的诊断价值。方法:回顾性分析196例经手术病理证实为T1期非小细胞肺癌患者的动脉期CT图像及临床病理资料,其中淋巴结转移患者75例,未淋巴结转移患者121例,将患者按照7∶3的比例随机分为训练组128例,测试组58例,通过ITK-SNAP勾画三维容积感兴趣区(VOI),Analysis-Kinetics软件外扩勾画好的VOI并提取三维影像组学特征,使用Feature Explorer软件进行特征分析,然后建立仅包含瘤内影像组学标签、包含肿瘤周围3 mm组织CT图的影像组学标签分别结合临床病理信息标签的联合模型。使用受试者操作特征曲线(ROC)分析评价模型的诊断效能。结果:瘤内组、含瘤周组的影像组学模型的AUC值分别为0.914、0.947,联合有统计学意义的临床及病理组织标签建立组合模型后,组合模型瘤内组及含瘤周组AUC值分别为0.938和0.967。结论:与仅瘤内影像组学模型相比,含3 mm瘤周组影像组学特征与临床病理指标的联合模型提高了T1期非小细胞肺癌淋巴结转移的预测能力。Objective:To investigate the diagnostic value of the combined model of radiomics features on CT images of intratumoral and peritumoral lung parenchyma and clinicopathological signatures in predicting lymph node metastasis of stage T1 non-small cell lung cancer.Methods:The arterial CT images and clinicopathological data of 196 patients with stage T1 non-small cell lung cancer confirmed by postoperative pathology were retrospectively analyzed,including 75 patients with lymph node metastasis and 121 patients without lymph node metastasis.Patients were randomly divided into the training group(128 cases)and the test group(58 cases)according to the ratio of 7∶3.VOIs were delineated by ITK-SNAP,and VOIs were expanded and radiomic features were extracted by Analysis-Kinetics.Feature explorer was used to analyze features and establish intratumoral model,combining intratumoral with peritumoral 3 mm tissue model and combined models.The receiver oper-ating characteristic(ROC)curve was used to evaluate the diagnostic effectiveness of the models.Results:The AUC value of the intratumoral model was 0.914 and the AUC value of combining intratumoral with peritumoral 3 mm tissue model was0.947.Combining clinicopathological signatures,the combined models were 0.938 and 0.967,respectively.Conclusion:The combined model of 3 mm peritumoral radiomic features and clinicopathological signatures improved the predictive ability of lymph node metastasis in stage T1 non-small cell lung cancer compared with the intratumoral radiomic model alone.

关 键 词：癌 非小细胞肺 体层摄影术 螺旋计算机 

分 类 号：R734.2[医药卫生—肿瘤] R814.42[医药卫生—临床医学]