丛梦迪,丛力宁,张延伟,任嘉梁,李扬,徐同欣,李焱.基于CT影像组学预测临床ⅠA期老年非小细胞肺癌淋巴结转移[J].中国介入影像与治疗学,2021,18(2):95-99
基于CT影像组学预测临床ⅠA期老年非小细胞肺癌淋巴结转移
Radiomics based on CT for predicting lymph node metastases in elderly patients with clinical stage ⅠA non-small cell lung cancer
投稿时间:2020-07-31  修订日期:2020-12-06
DOI:10.13929/j.issn.1672-8475.2021.02.008
中文关键词:  肺肿瘤  淋巴结  肿瘤转移  体层摄影术,X线计算机  影像组学
英文关键词:lung neoplasms  lymph nodes  neoplasm metastasis  tomography, X-ray computed  radiomics
基金项目:河北省医学科学研究计划项目(20200659)。
作者单位E-mail
丛梦迪 河北省儿童医院CT磁共振科, 河北 石家庄 050031  
丛力宁 河北省儿童医院放射科, 河北 石家庄 050031 112029279@qq.com 
张延伟 广州中医药大学第三附属医院放射科, 广东 广州 510006  
任嘉梁 通用电气药业有限公司, 北京 100176  
李扬 河北医科大学第四医院CT磁共振科, 河北 石家庄 050010  
徐同欣 河北医科大学第四医院CT磁共振科, 河北 石家庄 050010  
李焱 河北省儿童医院CT磁共振科, 河北 石家庄 050031  
摘要点击次数: 129
全文下载次数: 69
中文摘要:
      目的 探讨基于胸部增强CT影像组学模型预测临床ⅠA期非小细胞肺癌(NSCLC)老年患者淋巴结转移(LNM)的价值。方法 回顾性分析361例临床ⅠA期NSCLC老年患者术前增强CT及临床资料。术后病理显示其中87例LNM(LNM组)、274例无LNM(无LNM组),比较2组临床及影像学表现差异。提取术前增强CT影像学特征,进行归一化和降维,采用最小绝对收缩选择算子(LASSO)法筛选最优影像组学特征,建立影像组学模型。按7∶3比例将患者分为训练组和测试组,于训练组中以10次交叉验证法获得最佳影像组学预测模型。根据临床ⅠA期NSCLC老年患者LNM影响因素建立LNM临床预测模型,以之预测训练组和测试组LNM,并以ROC曲线评价2种模型对于训练组和测试组的诊断效能。结果 共于所有病灶中提取396个影像组学特征,经归一化后采用LASSO法获得5个最佳影像组学特征建立影像组学模型,并获得最佳影像组学模型,以之预测训练组和测试组LNM的AUC值分别为0.816和0.797,均高于临床模型(0.650和0.686,P均<0.05)。结论 基于胸部增强CT的影像组学模型可用于预测临床ⅠA期NSCLC老年患者LNM。
英文摘要:
      Objective To investigate the value of radiomics model based on enhanced CT for predicting lymph node metastasis (LNM) in elderly patients with clinical stage ⅠA non-small cell lung cancer (NSCLC). Methods Clinical and preoperative enhanced CT data of 361 elderly patients with clinical stage ⅠA NSCLC were retrospectively analyzed. Pathological results showed 87 patients with LNM (LNM group) and 274 without LNM (non-LNM group). The clinical and CT manifestations of patients with and without LNM were analyzed and compared between groups. The imaging features were extracted from preoperative enhanced CT and then normalized, and dimensionality reduction were performed. The least absolute shrinkage and selection operator (LASSO) method was used to select the optimal radiomics features to establish the radiomics model. Then all the patients were divided into training group and testing group in a ratio of 7∶3, and 10 cross-validation method was used to obtain the best radiomics predicted model in training group. The clinical model was established by using the clinical factors influencing LNM of elderly patients with clinical stage ⅠA NSCLC. Clinical model and radiomics model were used to predict the LNM of patients in training group and testing group, respectively, and ROC curve method was used to evaluate the diagnostic efficacy of the two models in training group and testing group. Results Totally 396 radiomics features were extracted. After normalization, 5 optimal radiomics features were obtained by using LASSO method to establish the radiomics model, and the optimal imaging radiomics model was obtained. AUC of the radiomics model for predicting LNM in training group and testing group was 0.816 and 0.797, respectively, higher than those of the clinical model (0.650, 0.686, both P<0.05). Conclusion Radiomics model based on enhanced chest CT could be used to predict LNM in elderly patients with clinical stage ⅠA NSCLC.
查看全文  查看/发表评论  下载PDF阅读器
关闭