【摘要】 目的 探讨含血清标志物-CT影像组学列线图模型诊断非小细胞肺癌(NSCLC)患者纵隔淋巴结转移的价值。方法 选取2021年1月至2023年12月就诊的163例NSCLC患者作为训练集，根据术后病理分为纵隔淋巴结转移组和非转移组，另选取2024年1至4月就诊的43例NSCLC患者作为验证集。统计训练集一般资料、血清标志物、CT图像，采用Double-Z软件提取影像组学特征，Lasso算法筛选最佳特征子集，多因素logistic回归分析NSCLC患者纵隔淋巴结转移的影响因素，构建血清标志物-CT影像组学列线图模型，受试者工作特征曲线、校准曲线、决策曲线分析列线图模型在训练集及验证集中的诊断效能。结果 转移组癌胚抗原(CEA)、糖类抗原125(CA125)、细胞角蛋白19片段抗原21-1(CYFRA21-1)、CT影像组学评分、肿瘤分期Ⅲ期占比及有吸烟史占比高于非转移组(P<0.05)。CEA、CA125、CYFRA21-1、肿瘤分期、吸烟史及CT影像组学评分是NSCLC患者纵隔淋巴结转移独立危险因素(P<0.05)。含血清标志物-CT影像组学列线图模型在训练集和验证集中诊断NSCLC患者纵隔淋巴结转移的曲线下面积分别为0.835、0.891,诊断结果与实际结果高度一致，在0.1～0.7、0.0～0.8范围内净收益率最高。结论 含血清标志物-CT影像组学列线图模型诊断NSCLC患者纵隔淋巴结转移效能高，临床实用性强，有助于指导临床诊治。更多还原

【Abstract】 Objective To explore the diagnostic value of serum marker-CT radiomics nomogram model for mediastinal lymph node metastases(MLNM) in patients with non-small cell lung cancer(NSCLC). Methods A total of 163 NSCLC patients treated from January 2021 to December 2023 were selected as the training set and divided into MLNM group and non-MLNM group according to postoperative pathology. Another 43 NSCLC patients treated from January to April 2024 were selected as the validation set. The general data, serum markers and CT images in the training set were recorded. Double-Z software was used to extract the radiomics features, and Lasso algorithm was used to screen the optimal feature subset. The influencing factors of MLNM in NSCLC patients were analyzed by multivariate logistic regression. The serum marker-CT radiomics nomogram model was constructed, and the diagnostic efficiency of this nomogram model in the training set and validation set was analyzed by drawing the receiver operating characteristic(ROC) curve, calibration curve and decision curve. Results Carcinoembryonic antigen(CEA), carbohydrate antigen 125(CA125), cytokeratin 19 fragment antigen 21-1(CYFRA21-1), CT radiomics score, the proportion of tumor stage III and smoking history in MLNM group were higher than those in non-MLNM group(P<0.05). CEA, CA125, CYFRA21-1, tumor stage, smoking history and CT radiomics score were independent risk factors for MLNM in NSCLC patients(P<0.05). In the training set and validation set, the area under the ROC curve of the serum marker-CT radiomics nomogram model for diagnosing MLNM in NSCLC patients was 0.835 and 0.891, respectively. The diagnosis results were highly consistent with the actual results, and the net return rate was the highest in the range of 0.1-0.7 and 0.0-0.8. Conclusion The diagnosis of MLNM in NSCLC patients with serum marker-CT radiomics nomogram model has high efficacy and strong clinical practicability, which is helpful to guide clinical diagnosis and treatment.更多还原