摘要
目的 探讨一阶组学特征分析在钆对比剂脑内沉积评估中的价值及最佳量化参数.方法 回顾性分析2010年1月至2021年12月收集的114例至少行4次线性钆对比剂MR增强检查的患者影像,测量患者每次检查平扫T1WI图像齿状核/脑桥信号强度比值,使用Spearman秩相关性分析评估信号强度比值和增强次数之间的关系;绘制首次和末次M RI平扫T1WI的齿状核/脑桥信号强度比值ROC曲线,并计算其曲线下面积AUC值,评估其诊断效能.采用ITK-SNAP软件,分别在首次及末次平扫T1WI图像上勾画双侧齿状核的ROI,提取一阶组学特征,绘制首次和末次检查有统计学差异特征的ROC曲线,计算其AUC值,评估筛选后的一阶组学特征的诊断效能.并利用Delong方法比较筛选后的一阶组学特征与信号强度比值的诊断效能.结果 多次行线性钆对比剂检查增强次数与平扫T1WI图像齿状核/脑桥信号强度比值之间存在正相关关系(r=0.570).齿状核/脑桥信号强度比值AUC为0.839.一阶组学特征中能量的AUC值(0.992)最高并高于信号强度比值(Z=5.6091,P<0.001).结论 一阶组学特征中的能量、总能量在钆对比剂脑内沉积评估中的诊断效能优于齿状核/脑桥信号强度比值,其中能量具有最高的诊断效能.
Abstract
Objective To investigate the value and optimal quantitative parameters of first-order omics feature analysis in the evaluation of gadolinium contrast agent deposition in the brain.Methods The images of 114 patients who underwent at least four linear gadolinium contrast agent MR enhancement examinations were retrospectively analyzed.The ratio of dentate nucleus/pons signal intensity in T1WI images of each examination was measured.Spearman rank correlation analysis was used to evaluate the relationship between signal intensity ratio and enhancement times.The receiver operating characteristic(ROC)curves of the dentate nucleus/pons signal intensity ratio of the first and last MRI unscanned T1WI were drawn,and the area under thecurve(AUC)was calculated to evaluate its diagnostic efficacy.Using ITK-SNAP software,the ROIs of bilateral dentate nucleus were delineated on the first and last unscanned T1WI images,and the first-order omics features were extracted.The ROC curves with statistically significant differences between the first and last examination were drawn,and the AUC values were calculated to evaluate the diagnostic efficacy of the screened first-order omics features.Delong method was used to compare the diagnostic efficacy of the screened first-order omics features with the ratio of signal intensity.Results There was a positive correlation between the intensity ratio of dentate nucleus/pons on plain T1WI images(r=0.570).The signal intensity ratio of dentate nucleus to pons(AUC)was 0.839.The AUC value of energy(0.992)was the highest among the first-order omics features and higher than the signal intensity ratio(P<0.001).Conclusion The diagnostic efficacy of first-order omics features of energy and total energy in the evaluation of gadolinium contrast agent intracerebral deposition is better than that of dentate nucleus/pons signal intensity ratio,and energy has the highest diagnostic efficacy.