【摘要】 目的 探讨动态增强扫描CT征象对原发性腹膜后肿瘤良恶性鉴别诊断价值。方法 选取322例经病理证实为原发性腹膜后肿瘤患者CT增强检查资料，对临床资料与影像学征象进行单因素分析，对差异有统计学意义的参数行Logistic回归分析，基于全部参数建立模型1（综合模型），基于内部成分与强化特征建立模型2（密度模型），基于形态学特征建立模型3（形态学模型），分析评价模型预测价值。结果 综合模型对腹膜后肿瘤良恶性预测性能最佳，其敏感度、特异度、准确度分别为81.3%、86.4%、83.2%,ROC曲线下面积为0.905。预测准确度及AUC高于密度模型（74.5%和0.799）和形态学模型（75.8%和0.849）。结论 综合临床与CT增强征象有利于提高良恶性肿瘤鉴别能力。鉴别腹膜后肿瘤良恶性的最佳参数组合为性别、不明显-轻度强化、脂肪密度、周围肿大淋巴结、周围器官推压以及血管侵犯。更多还原

【Abstract】 Objective To investigate the value of dynamic contrast-enhanced CT in differentiating between benign and malignant primary tumors of the retroperitoneum. Methods We retrospectively collected data of 322 patients with pathologically confirmed retroperitoneal tumor on CT enhancement examination, performed univariate analysis between clinical data and imaging signs. The parameters with statistically significant differences were subjected to logistic regression analysis. Model 1(integrated model) was built based on all parameters, model 2(density model) based on internal component and enhancement features, and model 3(morphological model) based on morphological features. Finally, the model predictive value was evaluated.Results The integrated model showed the best prediction performance for benign and malignant retroperitoneal tumors, with sensitivity of 81.3%, specificity of 86.4%, accuracy of 83.2%, and area under ROC curve of 0.905. The prediction accuracy and AUC were higher than density model(74.5% and 0.799) and morphology model(75.8% and 0.849). Conclusion The integrated clinical data with CT enhanced signs is beneficial to improve the discrimination between benign and malignant tumors.The best combination of parameters to differentiate between benign and malignant retroperitoneal tumors is gender, inconspicuous-mild enhancement, fat density, enlarged peripheral lymph nodes, peripheral organ push pressure, and vascular invasion.更多还原