基于T2WI灰度共生矩阵在鉴别高低级别胶质瘤中的应用

T2WI Based Gray Level Co-occurrence Matrix in Application of Differential Diagnosis of Grade of Glioma

目的:探讨T2WI灰度共生矩阵参数在鉴别高低级别胶质瘤中的价值。方法:收集54例病理证实为胶质瘤患者,其中高级别胶质瘤31例,低级别胶质瘤27例。在T2WI图像上使用ImageJ软件手动勾画出肿瘤最大层面的ROI,采用基于Matlab编写的软件提出灰度共生矩阵相关纹理特征,包括相关、能量,逆差距和熵,并采用独立样本t检验比较高低级别胶质瘤各特征间的差异,对于有统计学意义的特征绘制受试者工作特征曲线(ROC),并计算界值相应的敏感性和特异性。结果:高级别胶质瘤的纹理特征逆差距有意义较低级别低(P<0.001),高低级别胶质瘤的纹理特征相关性,能量和熵无统计学差异(P=0.261-0.849)。逆差距的ROC曲线下面积为0.895,以0.972为界值鉴别高低级别胶质瘤的敏感性和特异性分别为93.8%和75.0%。结论:基于T2WI灰度共生矩阵可用于鉴别高低级别胶质瘤,逆差距是鉴别二者的重要指标。

Purpose： To investigate the values of T2WI based gray level co-occurrence matrix in differential diagnosis of low- and high-grade glioma. Methods： The MRI data of fifty-four patients with pathological proved glioma （31 patients with high-grade glioma and 27 with low-grade glioma） were collected in our study. Region of interest （ROI） was manually drawn on the largest cross section of T2WI of the tumor. Texture analysis parameters of correlation, energy, inversed differential moment （IDM） and entropy were calculated using in-house Matlab- based tool. Independent samples t-test was used to compare the difference of each parameter between high- and low-grade glioma. Then receiver operation curves （ROCs） of parameters with statistical differences were drawn and sensitivity, specificity were calculated. Results： IDM of high-grade glioma was lower than that of low-grade glioma （P〈0.001）, while correlation, energy and entropy were with no statistical difference （P=0.261-0.849）. The areas under ROC curve of IDM was 0.895. The sensitivity and specificity at the optimal cutoff value of 0.972 for differing high- and lower-grade tumor were 93.8% and 75.0%, respectively. Conclusion： Gray level co-occurrence matrix based on T2WI is valuable for differentiating high-grade from low-grade glioma, and IDM has the most value in the differentiation.