目的：探讨MRI图像纹理分析在小脑半球原发性脑淋巴瘤和单发转移瘤鉴别诊断中的应用价值。方 法：回顾性收集经手术病理证实的位于小脑半球的原发性中枢神经系统淋巴瘤（PCNSL）20例和实性单发 转移瘤22例的增强MRI图像。采用纹理分析方法在增强T1WI轴位图像上获取整个瘤体的平均值、中位 数、标准差、峰度、熵值、偏度和异质性等定量参数，并进行统计学分析。结果：在纹理分析中，PCNSL组标 准差、熵值、偏度及异质性明显低于转移瘤组，而峰度明显高于转移瘤组（均P＜0.05）；异质性参数鉴别诊断 效能明显优于其它参数值，有较高的敏感性、特异性及准确性；对标准差、熵值、偏度、峰度及异质性5个纹 理分析定量参数进行多参数联合分析，其鉴别效能较利用单个纹理分析定量参数鉴别的效能高。结论：纹 理分析部分定量参数（标准差、熵值、偏度、峰度及异质性）有助于鉴别小脑半球PCNSL和单发转移瘤。

To investigate the value of MRI texture analysis in differentiating the diagnoses of primary cerebral nervous system lymphoma (PCNSL) and solitary metastasis in cerebellar hemispheres. Methods: We retrospectively analyzed the contrast-enhanced MRI of 20 patients with PCNSL and 22 patients with solitary solid metastasis with all cases having been confirmed by operative pathological examination. Applying MRI texture analysis to axial contrast-enhanced T1WI, we obtained the mean, median, standard deviation, kurtosis, entropy, skewedness, and inhomogeneity parameters and then statistically analyzed these quantitative parameters. Results: There were statistically significant differences in standard deviation, entropy, skewedness, kurtosis, and inhomogeneity between the two groups (all P<0.05). Inhomogeneity showed greater differential diagnostic efficacy than other parameters and demonstrated greater sensitivity, specificity, and accuracy. When applying multiparameter analysis for differential diagnosis using standard deviation, entropy, skewedness, kurtosis, and inhomogeneity obtained from MRI texture analysis, the diagnostic efficacy was superior to that when using individual parameters alone in analysis. Conclusion: Some of the quantitative parameters from MRI texture analysis (standard deviation, entropy, skewedness, kurtosis, and inhomogeneity) could be used in the differential diagnosis of PCNSL and solitary metastasis in cerebellar hemispheres.