临床-影像组学列线图术前对Ⅰ~ⅢA期原发性肺腺癌N_(1)~N_(2)淋巴结转移的预测价值  

The value of clinical-radiomics nomogram in preoperative predicting N1-N2 lymph node metastasis in patients with stageⅠto Ⅲ A primary lung adenocarcinoma

作　　者：徐鹤 谢宗玉 徐鹏 李磊 赵剑宇 王礼俊 周辉 XU He;XIE Zongyu;XU Peng;LI Lei;ZHAO Jianyu;WANG Lijun;ZHOU Hui(Department of Medical Imaging,Fengyang County People’s Hospital,Fengyang,Anhui Province 233100,China;Department of Radiology,the First Affiliated Hospital of Bengbu Medical University,Bengbu,Anhui Province 233000,China;Department of Medical Imaging,the People’s Hospital of Bozhou,Bozhou,Anhui Province 236800,China;Department of Thoracic Surgery,Fengyang County People’s Hospital,Fengyang,Anhui Province 233100,China)

机构地区：[1]凤阳县人民医院医学影像科,安徽凤阳233100 [2]蚌埠医科大学第一附属医院放射科,安徽蚌埠233000 [3]亳州市人民医院医学影像科,安徽亳州236800 [4]凤阳县人民医院胸外科,安徽凤阳233100

出　　处：《实用放射学杂志》2024年第8期1253-1258,共6页Journal of Practical Radiology

基　　金：安徽省重点研究与开发计划项目(2022e07020033);滁州市科技计划项目(农村与社会发展领域)(2022ZD007);凤阳县科技计划项目(SF2022-04)。

摘　　要：目的探讨临床-影像组学列线图术前对Ⅰ~ⅢA期原发性肺腺癌(PLA)N_(1)~N_(2)淋巴结转移的预测价值。方法将164例PLA患者分为训练集(n=114)和验证集(n=50),分别构建3种logistic回归预测模型。采用受试者工作特征(ROC)曲线及曲线下面积(AUC)评价模型的预测效能,模型间AUC差异采用DeLong检验,校准曲线评价校准度,决策曲线分析评价临床获益性。结果影像组学模型由8个影像组学特征[即影像组学评分(Radscore)]构成;肿瘤类型(中央型或周围型)与肿瘤大小构成临床模型;肿瘤类型、肿瘤大小以及Radscore构成列线图模型。训练集中,列线图模型预测效能(AUC=0.909)高于临床模型(AUC=0.748)和影像组学模型(AUC=0.814),AUC差异均有统计学意义(P<0.05)。验证集中,列线图模型预测效能(AUC=0.875)高于临床模型(AUC=0.682),AUC差异有统计学意义(P<0.05)。列线图模型预测效能亦高于影像组学模型(AUC=0.799),AUC差异无统计学意义(P>0.05)。校准曲线显示临床-影像组学列线图具有良好的校准度,决策曲线显示出良好的临床获益性。结论临床-影像组学列线图术前对Ⅰ~ⅢA期PLA N_(1)~N_(2)淋巴结转移具有良好的预测能力,明显高于单一的影像组学和临床因素。Objective To investigate the preoperative predictive value of clinical-radiomics nomogram on N1-N2 lymph node metastasis in patients with stageⅠtoⅢA primary lung adenocarcinoma(PLA).Methods A total of 164 PLA patients were divided into a training set(n=114)and an validation set(n=50).Three logistic regression models were created separately and the predictive efficacy of the model was evaluated using the receiver operating characteristic(ROC)curve and area under the curve(AUC)respectively.The AUC difference between models was tested by the DeLong test.The calibration degree was evaluated by the calibration curve.Decision curve analysis was performed to evaluate the benefits of clinical application.Results The radiomics model consisted of 8 imaging features[Radiomics score(Radscore)].The clinical model was composed of tumor type(central or peripheral)and tumor size.The tumor type,tumor size and Radscore formed the nomogram model.In the training set,the prediction of the nomogram model was more effective(AUC=0.909)than the clinical model(AUC=0.748)and the radiomics model(AUC=0.814),while the differences in AUC were statistically significant(P<0.05).In the validation set,the prediction of the nomogram model was more effective(AUC=0.875)than the clinical model(AUC=0.682),and the difference in AUC was statistically significant(P<0.05).The prediction of the nomogram model was also more effective than the radiomics model(AUC=0.799),but the difference in AUC was not statistically significant(P>0.05).The calibration curve showed that the clinical-radiomics nomogram had a high level of calibration and the decision curve analysis showed good benefits from clinical application.Conclusion The clinical-radiomics nomogram is proven to be more effective than radiomics or clinical factors alone in the preoperative prediction of stageⅠtoⅢA PLA N1-N2 lymph node metastasis.

关 键 词：临床-影像组学列线图 原发性肺腺癌 淋巴结转移 

分 类 号：R445[医药卫生—影像医学与核医学] R734.2[医药卫生—诊断学]