影像组学在预测非小细胞肺癌分子标志物P63中的应用价值

目的：建立基于非小细胞肺癌(non-small cell lung cancer，NSCLC)肿瘤CT图像的影像组学模型，预测NSCLC分子标志物P63的表达状态。方法：回顾性分析2014年1月至2018年3月接受CT扫描的245例NSCLC患者。患者均经组织病理学检查确诊，并在CT检查后2周内进行P63表达状态检测。通过MaZda软件提取CT平扫图像的影像组学特征，并且定义肿瘤CT图像的主观影像征象。使用Lasso-logistic回归模型进行特征筛选并分别建立影像组学模型、主观影像征象模型及融合诊断模型。通过受试者操作特征(receiver operator characteristic，ROC)曲线评估每个模型的预测性能，并采用Delong检验进行比较。结果：在245例患者中，P63阳性96例，P63阴性149例。主观影像征象模型由6个影像征象组成。通过特征选择，影像组学模型包括8个影像组学特征。主观影像征象模型和影像组学模型预测P63表达状态的ROC曲线下面积分别为0.700和0.755，二者差异无统计学意义(P>0.05)。融合诊断模型较另2种模型具有最佳预测能力，ROC曲线下面积为 0.817(P<0.01)。结论：基于CT图像的影像组学模型可以预测NSCLC分子标志物P63的表达状态；融合影像组学特征和主观影像征象的诊断模型可以显著提高模型的预测性能，有助于无创性了解肺癌细胞分子水平信息。

Radiomics in predicting tumor molecular marker P63 for non-small cell lung cancer

Objective: To establish a radiomics signature based on CT images of non-small cell lung cancer(NSCLC) to predict the expression of molecular marker P63.Methods: A total of 245 NSCLC patients who underwent CT scans were retrospectively included.All patients were confi rmed by histopathological examinations and P63 expression were examinedwithin 2 weeks aft er CT examination. Radiomics features were extracted by MaZda soft ware andsubjective image features were defi ned from original non-enhanced CT images. Th e Lasso-logisticregression model was used to select features and develop radiomics signature, subjective imagefeatures model, and combined diagnostic model. Th e predictive performance of each model wasevaluated by the receiver operating characteristic (ROC) curve, and compared with Delong test.Results: Of the 245 patients, 96 were P63 positive and 149 were P63 negative. The subjectiveimage feature model consisted of 6 image features. Through feature selection, the radiomicssignature consisted of 8 radiomics features. Th e area under the ROC curves of the subjective imagefeature model and the radiomics signature in predicting P63 expression statue were 0.700 and 0.755,respectively, without a signifi cant diff erence (P>0.05). Th e combined diagnostic model showed thebest predictive power (AUC=0.817, P<0.01).Conclusion: The radiomics-based CT scan images can predict the expression status of NSCLCmolecular marker P63. Th e combination of the radiomics features and subjective image features cansignifi cantly improve the predictive performance of the predictive model, which may be helpful toprovide a non-invasive way for understanding the molecular information for lung cancer cells.