论文部分内容阅读
目的探讨三维非刚性配准在提高肝脏动态增强MR应用中的可重复性。方法前瞻性分析18例行动态增强MRI检查的肝硬化背景肝细胞肝癌患者的资料。使用Omni Kinetics软件对获得的图像进行三维非刚性配准,采用Extended Tofts双输入模型对配准前、后的图像进行计算。由两位观测者采用盲法测量配准前、后同层面病灶、肝脏及竖脊肌的K~(trans)、K_(ep)值,并对测量值的变异度、一致性进行组间(两位观测者之间)及组内(一位观测者间隔一周测量两次)比较,并使用Bland-Altman Plot对组间和组内测量值的可重复性进行分析。结果 18例三维非刚性配准后同层不同期两幅图像有差异,像素均值为32761±13575,小于配准前(43202±20884),差异有统计学意义(t=2.637,P<0.05)。相对差值百分比箱图显示组间及组内病灶、肝脏、竖脊肌K~(trans)、K_(ep)均数配准后差异均小于配准前,离群值及极值减少或消失;配准后组内竖脊肌K~(trans)和K_(ep)的四分位间距稍增大,余四分位间距均变小;配准后组间四分位间距均变小。配准后组内、组间病灶、肝脏、竖脊肌K~(trans)、K_(ep)可重复性均提高,均值差变小。结论三维非刚性配准可提高病灶、肝脏及竖脊肌K~(trans)、K_(ep)的可重复性。
Objective To investigate the reproducibility of three-dimensional non-rigid registration in improving the dynamic enhancement of liver MR. Methods Prospective analysis of 18 patients with liver cirrhosis with hepatocellular carcinoma after dynamic contrast-enhanced MRI. The obtained images were three-dimensionally non-rigidly registered using the Omni Kinetics software and the images before and after registration were calculated using the Extended Tofts dual-input model. Two observers used blind method to measure K trans and K ep values of lesion, liver and erector spinae before and after registration, and the variability and consistency of the measured values between groups Between two observers) and within the group (one observer was measured twice a week apart) and the repeatability of the intergroup and in-group measurements was analyzed using Bland-Altman Plot. Results There were significant differences between the two images at 18 different stages after the three-dimensional non-rigid registration. The pixel mean was 32761 ± 13575, which was lower than that before registration (43202 ± 20884), the difference was statistically significant (t = 2.637, P <0.05) . The relative difference percentage box diagram showed that the differences of K trans and K ep were less than those before registration and the outliers and extremities decreased or disappeared After registration, the interquartile range of trans and k ep of epinephrine increased slightly and the interquartile range became smaller. The interquartile range of inter-group became smaller after registration. Within the group after registration, the repeatability of K trans and K ep of the lesion, liver and erector spinae increased and the mean difference became smaller. Conclusion Three-dimensional non-rigid registration can improve the repeatability of K trans and K ep of lesions, liver and erector spinae.