【摘要】 目的：探讨CT平扫图像纹理分析鉴别骨巨细胞瘤（Giant cell tumor of bone,GCT）与动脉瘤样骨囊肿（Aneurismal bone cyst,ABC）的价值。方法：回顾性分析经手术病理证实的14例骨巨细胞瘤和15例动脉瘤样骨囊肿的影像学资料。在CT图像上手动勾画ROI测量CT值及提取纹理特征参数。采用两独立样本t检验或MannWhitney U检验比较两组间CT值及纹理参数的差异，对得到的CT值及纹理参数进行单因素和多因素Logistic回归分析，绘制ROC评价CT值、纹理参数和纹理参数模型的诊断效能。结果：两组间CT值差异有统计学意义，鉴别GCT和ABC的AUC为0.162。CT平扫图像共提取1 044个纹理参数，经筛选获得8个有统计学意义的参数（orrelation＿angle45＿offset4、 GLCMEntropy＿AllDirection＿offset1、 sumVariance、 sumEntropy、 histogramEntropy、GLCMEntropy＿angle45＿offset1、Inertia＿angle0＿offset2、ShortRunEmphasis＿angle135＿offset6），其鉴别GCT和ABC的AUC分别为0.343、0.843、0.776、0.800、0.800、0.848、0.795、0.771；多因素Logistic回归分析获得的纹理参数模型预测GCT和ABC的AUC为0.900,CT值结合纹理参数获得的模型鉴别GCT和ABC的AUC为0.948。结论：CT值结合纹理参数模型对于鉴别GCT和ABC具有较高价值。更多还原

【Abstract】 Objective : To explore the value of CT texture analysis in differential diagnosis of giant cell tumor of bone(GCT) from aneurismal bone cyst(ABC). Methods : A total of 14 cases with GCT and 15 with ABC confirmed by operation and pathology with CT examinations were retrospectively analyzed. On the CT image, manually delineated the ROI to measure the density of the CT values and extracted the texture parameters. Two independent sample t-test or Mann-Whitney U test were used to compare the difference between CT and texture parameters. The obtained CT values and texture parameters were analyzed by univariate logistic regression and multivariate Logistic regression analysis was used. A receiver operating characteristic curve was performed to evaluate diagnostic performance of the CT values, texture parameter and texture parameter model for differential diagnosis of GCT from ABC. Results : There was a statistically significant difference in CT values between GCT and ABC. The AUC of the CT values for differential diagnosis of GCT from ABC was 0. 162. A total of 1 044 texture parameters were extracted from the CT scan image, with 8 statistically significant parameters(Correlation＿angle45＿offset4, GLCMEntropy＿AllDirection＿offset1, sumVariance, sumEntropy, histogramEntropy, GLCMEntropy＿angle45＿offset1, Inertia＿angle0＿offset2, ShortRunEmphasis＿angle135＿offset6), AUC values of 8 parameters for differentiation of GCT from ABC were 0. 343, 0. 843, 0. 776, 0. 800, 0. 800, 0. 848, 0. 795, 0. 771, respectively. Multivariate Logistic regression analysis showed that diagnostic accuracies for differentiation of GCT from ABC were 90. 00% using texture analysis predictors exclusively, and 94. 80% using a combined model of CT values and texture analysis predictors, respectively. Conclusion : The model obtained by CT value combined with texture parameters has a high value for differentiating GCT from ABC.更多还原