探讨增强MRI序列纹理分析在鉴别眼眶淋巴瘤和炎性假瘤中的诊断价值。

回顾性分析2013年1月至2019年4月于苏北人民医院经病理证实的眼眶淋巴瘤25例和炎性假瘤24例，术前进行MRI常规序列和脂肪抑制T1WI增强扫描，选取每例患者增强图像上肿瘤的最大层面区域，应用Mazda软件勾画肿瘤感兴趣区并计算灰度直方图纹理参数。提取出9个特征参数分别为：均值(mean)、方差(variance)、偏度(skewness)、峰度(kurtosis)和第1、10、50、90、99百分位数(Pere.1%、Pere.10%、Pere.50%、Pere.90%、Pere.99%)，分别比较淋巴瘤和炎性假瘤各个纹理参数之间的差异，其中符合正态分布的参数采用双样本t检验，不符合正态分布的参数采用秩和检验，并采用受试者工作特征曲线(receiver operaling characteristic curve，ROC)分析曲线下面积(area under curve，AUC)、敏感度、特异度。运用多变量Logistic回归分析对有统计学意义的纹理参数进行建模并绘制ROC曲线评价模型效能。

2种肿瘤在均值、方差、第50、90、99个百分位数参数中差异有统计学意义(P<0.05)。在偏度、峰度和第1百分位数参数中差异无统计学意义(P>0.05)。2组间纹理参数方差在阈值为113.20时具有最佳诊断效能，对应的敏感度、特异度、AUC分别是95.83%、76.00%、0.930(54.87%～90.64%)。

增强MRI序列纹理分析可以用于眼眶淋巴瘤和炎性假瘤的鉴别诊断，且参数方差具有最佳诊断效能。

To evaluate the value of enhanced magnetic resonance imaging(MRI)sequence texture analysis in the differential diagnosis of orbital lymphoma and inflammatory pseudotumor.

Retrospective analysis was performed on 25 cases of orbital lymphoma and 24 cases of inflammatory pseudotumor confirmed by pathology in Northern Jiangsu People′s Hospital from January 2013 to April 2019.Routine MRI sequence and enhanced fat suppression T1WI scan were performed before surgery.The maximum tumor layer area on the enhanced image of each patient was selected, and Mazda software was used to outline the tumor region of interest(ROI)and calculate the grayscale histogram texture parameters.Nine feature parameters were extracted, which were respectively: mean, variance, skewness, kurtosis, and Pere.1%, Pere.10%, Pere.50%, Pere.90%, Pere.99%.The differences of various texture parameters between lymphoma and inflammatory pseudotumor were compared, among which two-sample t test was used for parameters conforming to normal distribution, rank sum test was used for parameters not conforming to normal distribution, and receiver operating characteristic(ROC)curve was used to analyze area under curve(AUC), sensitivity and specificity.Multivariate logistic regression analysis was used to model statistically significant texture parameters and draw ROC curve to evaluate the model effectiveness.

There were statistically significant differences between the two tumors in mean, variance, Pere.50%, Pere.90%, Pere.99%(P<0.05). The differences in skewness, kurtosis and Pere.1% were not statistically significant(P>0.05). When the variance of texture parameters between the two groups was 113.20, it had the best diagnostic performance, and the corresponding sensitivity, specificity and AUC were 95.83%, 76.00% and 0.930(54.87%-90.64%), respectively.

Enhanced MRI sequence texture analysis can be used in the differential diagnosis of orbital lymphoma and inflammatory pseudotumor, and the parameter variance has the best diagnostic efficiency.