根据2013版肝脏影像报告与数据管理系统诊断肝脏病灶的一致性

目的观察根据2013版肝脏影像报告与数据管理系统LI-RADS(v2013)]诊断肝细胞癌(HCC)发病高风险人群肝脏病灶的一致性。方法由2名医师(医师1、2)独立双盲阅读813例有HCC发病风险患者的1115个病灶的CT/MRI资料,分别采用LI-RADS(v2013)、LI-RADS(v2012,第1版)即LI-RADS(v1.0)]进行诊断。分析医师间根据LI-RADS(v2013)评估病灶直径、主要影像学征象及分级诊断结果的一致性;以病理结果为金标准,比较根据LI-RADS(v2013)与LI-RADS(v1.0)进行诊断的结果差异。结果医师1、2评估病灶直径的组内相关系数(ICC)为0.998;评价动脉期高强化、门静脉期/延迟期低强化及包膜的Kappa系数分别为0.87、0.86、0.72;LI-RADS(v2013)分级诊断观察者内Kappa系数为0.85。2名医师协商后的最终分级诊断结果与金标准的Kappa系数为0.85;诊断阳性(LR3~5级)病灶的敏感度、特异度、阳性预测值(PPV)及阴性预测值(NPV)分别为100%(757/757)、80.17%(287/358)、91.43%(757/828)及100%(287/287)。根据LI-RADS(v2013)、LI-RADS(v1.0)诊断阳性病灶差异有统计学意义(P<0.001)。结论医师间评价LI-RADS(v2013)主要影像学征象及分级诊断结果的一致性均较好。

Consistency of diagnosing hepatic lesions according to LI-RADS (v2013)

Objective To observe the consistency of diagnosing liver lesions according to liver imaging reporting and data system of 2013 (LI-RADSv2013]) in people with high risk of hepatocellular carcinoma (HCC). Methods CT/MRI data of 813 patients (1 115 liver lesions) with high risk of HCC were double-blindly reviewed by 2 physicians independently, and LI-RADS (v2013) and LI-RADS (v1.0) were used for qualitative diagnosis of lesions. The consistency of lesion diameter, three major imaging features and grading diagnosis results based on LI-RADS (v2013) were evaluated between 2 observers. Pathological results were taken as the gold standards, and the diagnosis results obtained according to LI-RADS (v2013) and LI-RADS (v1.0) were compared. Results Intraclass correlation coefficient (ICC) of lesion diameter evaluated by physician 1 and 2 was 0.998, while the Kappa coefficient of arterial phase hyperenhancement, portal venous phase/later phase hypoenhancement and substantial for capsule was 0.87, 0.86 and 0.72, respectively. The Kappa coefficient of interobserver agreement of grading diagnosis using LI-RADS (v2013) was 0.85, same as that of final grading diagnosis after consultation of physicians and gold standards. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for diagnosing positive (LR3-5) lesions was 100% (757/757), 80.17% (287/358), 91.43% (757/828) and 100% (287/287), respectively. There was significant difference in diagnosis of positive lesions between LI-RADS (2013) and LI-RADS (v1.0) (P<0.001). Conclusion The intraobserver consistency of diagnosing liver lesions with major imaging features and grading of lesions using LI-RADS (v2013) are both good.