乙肝肝硬化背景下肝细胞肝癌的MRI表现:LI-RADS(2014版)定义征象的识别率分析

Recognition of the MR imaging features of HCC foci in cirrhotic liver:based on LI-RADS(2014)definition

目的:分析肝硬化背景上肝细胞肝癌(HCC)的MRI动态增强表现,探讨2014版肝脏影像报告及数据系统(LI-RADS)所定义的各种征象在MRI上的识别率。方法:2008年12月-2014年8月共54例乙肝肝硬化患者入组本多中心研究,所有患者均行MRI平扫和增强扫描、且经病理诊断证实有HCC。由两位阅片者对MR图像进行独立盲法读片,阅片方法遵循LI-RADS流程:对主要征象、次要征象进行评价(出现/未出现),计算2位阅片者对LI-RADS所定义的各种MRI征象的识别率。通过Kappa检验来分析两位阅片者对征象识别的一致性。结果:两位阅片者对病灶的3个主要征象(动脉期高强化、"廓清"表现、"包膜"表现)的识别率依次分别为83.3%和85.2%、77.8%和64.8%、51.9%和61.1%,阅片者间的一致性(Kappa值)分别为0.791、0.512和0.589。对于次要征象,大部分征象均有一定的识别率,以T2WI上稍高信号、扩散受限的识别率最高,分别为90.7%和87.0%、88.9%和90.7%,阅片者间的一致性(Kappa值)分别为0.813和0.899。结论:基于LI-RADS的诊断标准,平扫及动态增强MR图像上乙肝肝硬化背景上的HCC病灶的主要征象能较多地被识别,且阅片者的一致性较好;次要征象中T2WI稍高信号和扩散受限的识别率较高,阅片者间的一致性也非常好。

Objective：To evaluate the recognition rate of MR imaging features of hepatocellular carcinoma（HCC）in cirrhotic liver due to HBV based on criteria of LI-RADS（v2014）.Methods：From November 2008 to August 2014,54 patients with liver cirrhosis due to HBV infection were enrolled in this multicenter study.All the patients underwent pre-contrast and post-contrast MRI and DWI examination and were confirmed pathologically to have HCC foci.Two radiologists analyzed the images independently.The readers analyzed and categorized the features of the HCC foci based on the definition given by LI-RADS.The recognition rate was calculated for the major and ancillary imaging features.The Kappa test was used to assess inter-reader agreement for all descriptive variables.Results：The recognition rates of the major features of HCC by the two radiologists were as following：hyper-enhancement in arterial phase,83.3% and 85.2%（Kappa=0.791）; ＂ washout＂ appearance,77.8% and 64.8%（Kappa=0.512）; ＂capsule＂ appearance,51.9% and 61.1%（Kappa=0.589）.Most of the ancillary features could be observed,of which mild to moderate hyper-intensity on T2WI（90.7% and 87.0%,Kappa=0.813）and restricted diffusion（88.9% and 90.7%,Kappa=0.899）were more common than the other features.Conclusion：Based on LI-RADS definition,the major imaging features of HCC on MR images could be recognized with fairly good inter-reader agreement;in the ancillary imaging features,mild to moderate hyper-intensity on T2 WI and restricted diffusion might be recognized more frequently,with good inter-reader agreement.