【摘要】 目的：探讨基于CT平扫征象的预测模型，对肺梗死与胸膜下肺炎的鉴别诊断价值。方法：回顾性分析29例肺梗死与35例肺炎患者的胸部CT图像。对2组楔形征、血管征、反晕征、胸膜增厚、近心侧透光度减低征、肺体积缩小、胸腔积液、静脉分支可见征、支气管充气征等征象行单因素和多因素logistic回归分析，建立预测模型。采用ROC曲线分析预测模型的效能。结果：2组楔形征、血管征、反晕征、胸膜增厚、近心侧透光度减低征、肺体积缩小、胸腔积液、静脉分支可见征、多发差异均有统计学意义（均P<0.05）。静脉分支可见征多见于肺炎，余8个征象多见于肺梗死。多因素logistic回归分析显示，楔形征是肺梗死的独立预测因子，静脉分支可见征是肺梗死的保护因子。楔形征与静脉分支可见征构建的联合模型的AUC为0.925，高于血管征及楔形征单独诊断（均P<0.05）。结论：楔形征与静脉分支可见征构建的联合模型可较准确地鉴别肺梗死和肺炎。更多还原

【Abstract】 Objective :To explore the value of a predictive model of CT non-enhanced features in differential diagnosis between pulmonary infarction and pneumonia. Methods :Chest CT images of 29 patients with pulmonary infarction and 35 patients with pneumonia were analyzed retrospectively. Univariate and multivariate logistic regression analysis were performed to analyze the factors,including the wedge sign,vascular sign,reversed halo sign,pleural thickening,decreased proximal cardiac transmittance,lung volume reduction,pleural effusion,venous branch sign,and air bronchogram sign. And the predictive model was established. ROC curve was used to analyze the diagnostic efficacy. Results:There were statistically significant differences in the wedge sign,vascular sign,reversed halo sign,pleural thickening,decreased proximal cardiac transmittance,lung volume reduction,pleural effusion,venous branch sign(all P<0.05). The venous branch sign was more common in pneumonia,the other features were more common in pulmonary infarction. Multivariate logistic regression analysis showed that wedge sign was independent predictor of pulmonary infarction,while venous branch sign was protective factor of pneumonia. The AUC of the combined model constructed with wedge sign and venous branch sign reached 0.925,higher than that of wedge sign and vascular sign separately(both P<0.05). Conclusion:The combined model constructed with wedge sign and venous branch sign can accurately differentiate pulmonary infarction from pneumonia.更多还原