纹理分析在原发性脑淋巴瘤与高级别胶质瘤鉴别诊断中的应用价值

The Value of Texture Analysis in Differentiating Diagnosis Between Primary Cerebral Lymphoma and High-Grade Glioma

目的探讨MRI图像纹理分析对鉴别原发性脑淋巴瘤与高级别胶质瘤(HGG)的临床应用价值。方法回顾性分析经病理证实的15例原发性脑淋巴瘤与18例HGG患者的MRI表现,并对T1WI增强图像进行纹理分析,分别得到熵值、偏度、峰度及异质性等相关参数。结果原发性脑淋巴瘤好发于深部脑组织,如胼胝体区、额叶及基底节区,在MRI表现为T1WI呈等、低信号,T2WI呈等、稍高信号,边界较清,增强后多呈均匀明显强化;HGG在MRI表现为T1WI呈低信号,T2WI呈高信号,增强扫描后呈明显不均匀强化,病灶内部多伴有囊变、坏死及出血成分。原发性脑淋巴瘤与HGG比较,多发病灶、强化方式和累及中线特征在统计学上存在显著差异(P<0.05);纹理分析脑淋巴瘤组和HGG组中T1WI增强图像,脑淋巴瘤组的熵值与异质性参数低于HGG组的熵值与异质性参数,两者在统计学上存在显著差异(P<0.05)。同时,纹理分析异质性参数明显优于熵值;当异质性参数Cut-off点取0.175时,鉴别诊断脑淋巴瘤与HGG的敏感性、特异性和准确性分为86.7%、94.4%和90.9%。结论纹理分析的熵值、异质性参数结合常规MRI平扫和增强表现,对鉴别原发性脑淋巴瘤与HGG具有一定的临床应用价值。

Objective To investigate the value of texture analysis of MRI image in differentiating diagnosis between primary cerebral lymphoma and high-grade glioma（ HGG）. Methods The MRI imaging data of 14 cases of primary cerebral lymphoma and 18 cases of HGG with pathological proof were analyzed retrospectively. Texture analysis of T1 enhancement were performed to obtain entropy,skewness,kurtosis and inhomogeneity parameters. Results The most common site of primary cerebral lymphoma was deep brain tissue,such as the corpus callosum,the frontal lobe and the basal ganglia. Primary cerebral lymphoma usually showed equal or slight long T1 and T2signal on conventional MRI,enhancement was obvious and homogeneous. HGG showed low T1 and high T2 signal,enhancement was usually obvious and inhomogeneous with cystic degeneration,necrosis and hemorrhage. There is a statistically significant difference in the feature of multiple lesions,enhancement,crossing the midline,entropy and heterogeneity of texture analysis between primary cerebral lymphoma and HGG（ P 〉0. 05）. Tumor entropy and inhomogenity obtained in primary cerebral lymphoma group is lower than entropy and inhomogenity in HGG group on T1enhancement; And tumor inhomogenity is significantly better than tumor entropy parameter. When Cut-off point = 0. 175,tumor heterogeneity served as differentiating diagnostic criteria between primary cerebral lymphoma and HGG,the sensitivity was 86. 7%,the specificity 94. 4% and the accuracy 90. 9%. Conclusion It is of significant value to combine entropy and heterogeneity of texture analysis,conventional and contrast-enhanced MRI appearances in differentiating diagnosis between primary cerebral lymphoma and HGG.