基于常规MRI的纹理分析预测低级别胶质瘤1p/19q共缺失状态

Predicting the co-deletion of 1p/19q in low-grade glioma based on texture analysis of conventional MRI

目的探讨基于常规磁共振成像(MRI)纹理分析在预测低级别胶质瘤(LGG)1p/19q共缺失状态中的价值。方法回顾性纳入106例低级别胶质瘤患者,根据患者1p/19q共缺失状态分为2组,在不同MRI序列图像上勾画感兴趣区域(ROI),并提取ROI参数。基于单因素分析,使用受试者工作特征曲线(ROC)和多因素Logistics回归进行结果分析。结果ROC分析显示,T2WI偏度参数的灵敏度为92.9%,特异度为69.4%,AUC为0.857;T2WI变异系数的灵敏度为65.7%,特异度为69.4%,AUC为0.702。多因素logistics回归分析显示,T2WI(OR=1.004,95%CI:1.001~1.006,P=0.001)和CE-T1WI(OR=0.393,95%CI:0.206~0.748,P=0.004)偏度参数是低级别胶质瘤1p/19q共缺失状态的独立预测因素。结论基于常规MRI的纹理分析对预测低级别胶质瘤1p/19q共缺失状态具有重要价值。

Objective To explore the value of texture analysis based on conventional magnetic resonance imaging(MRI)in predicting 1p/19q co-deletion in low-grade glioma.Methods A total of 106 patients with low-grade glioma were retrospectively included.According to the co-deletion of 1p/19q,they were divided into two groups.The regions of interest(ROI)were outlined in MRI images,and the parameters of ROI were extracted.The results were analyzed by the receiver operating characteristic curve(ROC)and multivariate logistics regression.Results According to ROC analysis,the sensitivity of T 2WI skewness parameter was 92.9%,the specificity was 69.4%,and the AUC was 0.857.The sensitivity of T 2WI coefficient of variation was 65.7%,the specificity was 69.4%,and the AUC was 0.702.Multivariate logistics regression analysis indicated that T 2WI(OR=1.004,95%CI:1.001-1.006,P=0.001)and CE-T 1WI(OR=0.393,95%CI:0.206-0.748,P=0.004)skewness parameters were independent predictors of 1p/19q co-deletion in low-grade glioma.Conclusions Texture analysis based on conventional MRI can effectively predict the co-deletion of 1p/19q in low-grade glioma.