基于胸部CT平扫的影像组学模型预测肺癌隐匿性淋巴结转移的临床价值

The Clinical Value of Radiomics Model Based on Chest CT Plain Scan for Predicting Occult Lymph Node Metastasis in Lung Cancer

目的:建立基于胸部CT平扫的影像组学模型,探讨其在预测肺癌隐匿性淋巴结转移中的临床价值。方法:回顾性收集114例行胸部CT平扫的非小细胞肺癌患者影像及临床资料,共计125枚目标淋巴结纳入研究。采用Pyradiomics软件提取影像组学特征,通过最小冗余最大相关性和最小绝对收缩与选择算子分析获得组学模型。采用单因素方差分析对训练集和内部验证集患者的临床资料进行比较,采用受试者工作特征(ROC)曲线和决策曲线评估组学模型的临床价值。结果:经过筛选,共获得17个最具预测性的影像组学特征,并以此建立影像组学模型。训练集ROC曲线下面积(AUC)值为0.86,准确度、灵敏度和特异度分别为76.14%、91.30%、70.77%;内部验证集AUC值为0.85,准确度、灵敏度和特异度分别为70.27%、68.75%、71.43%,决策曲线显示模型具有较好的临床获益。结论:基于胸部CT平扫建立影像组学模型,其在预测肺癌OLM上具有较好的临床价值。

Purpose:To develop a radiomics model based on chest CT plain scan,and to evaluate its clinical value in predicting occult lymph node metastasis(OLM)of lung cancer.Methods:The imaging and clinical data of 114 patients with pathologically confirmed non-small cell lung cancer who underwent chest CT plain scanning were retrospectively analyzed,and 125 target lymph nodes were included in the study.Radiomics features were extracted by Pyradiomics software,and the radiomics model was obtained by minimum redundancy maximum relevance(mRMR)and least absolute shrinkage and selection operator(LASSO).The clinical data of patients in the training and internal validation cohort were compared by one-way ANOVA,and the clinical value of the radiomics model was evaluated by receiver operating characteristic(ROC)curve and decision curve.Results:After screening,17 most predictive radiomics features were obtained,and the radiomics model was established.The area under the curve(AUC)value of the training cohort was 0.86,with the accuracy,sensitivity and specificity of 76.14%,91.30%and 70.77%,respectively.The AUC value of the internal validation cohort was 0.85,with the accuracy,sensitivity and specificity of 70.27%,68.75%and 71.43%,respectively.The decision curve showed that the model had excellent clinical benefits.Conclusion:The radiomics model based on chest CT plain scan shows an excellent clinical value in predicting OLM of lung cancer.