目的 探讨超声造影肝脏影像报告与数据系统（CEUS LI-RADS）调整分类在≤ 5 cm肝细胞癌（hepatocellular carcinoma， HCC）中的诊断价值。 方法 回顾性分析461个肝癌高风险的肝脏结节CEUS资料，以病理结果为金标准。先参照2017版CEUS LI-RADS对结节进行分类，再根据以下两点进行调整：（1）动脉期高增强、开始廓清时间在45 ~ 60 s内的LR-M类结节重新判定为LR-5；（2）LR-4、LR-M类结节若合并AFP/DCP阳性则重新判定为LR-5。比较调整前后LR-5的诊断效能。 结果 379个HCC中，调整后LR-M类HCC占比从32.70%降至4.70%（χ2 = 97.367， P < 0.001），在肿瘤最大径总体组、≤ 3 cm组、3 ~ 5 cm组、0 ~ 2 cm组、2 ~ 3 cm组中，调整分类均显著提高灵敏度［（92.61%、92.41%、93.33%、90.30%、95.04%） vs.（62.27%、61.24%、65.56%、58.79%、64.46%）］、准确度［（90.5%、89.52%、91.84%、87.78%、91.60%） vs. （62.38%、60.38%、67.35%、58.89%、63.36%）］、AUC值［（0.766、0.742、0.842、0.752、0.726） vs.（0.630、0.586、0.765、0.594、0.572）］，阳性预测值达96%以上。 结论 CEUS LI-RADS调整分类在不降低阳性预测值的同时，显著提高了≤ 5 cm HCC诊断的灵敏度、准确度、AUC值，在≤ 2 cm病灶中同样适用，可为临床决策提供更可靠的参考依据。

Objective To explore the application value of the modified Contrast-Enhanced Ultrasound Liver Imaging Reporting and Data System （CEUS LI-RADS） in the diagnosis of hepatocellular carcinoma （HCC） with the diameter of ≤ 5 cm. Methods A retrospective analysis was performed on contrast-enhanced ultrasound （CEUS） data from 461 liver nodules in high-risk HCC patients， with pathological results serving as the gold standard. The nodules were classified according to the 2017 CEUS Liver Imaging Reporting and Data System （LI-RADS） and then modified based on the following two criteria： （1） nodules with arterial phase hyper-enhancement and a washout time of 45-60 s were re-classified as LR-5； （2） If LR-4 and LR-M nodules were combined with positive alpha-fetoprotein （AFP） or des-gamma-carboxy prothrombin （DCP）， they were re-judged as LR-5. The diagnostic efficacy of LR-5 before and after modification was compared. Results Among 379 HCC lesions， after modification， the proportion of HCC cases classified as LR-M decreased from 32.70% to 4.70% （χ2 = 97.367， P < 0.001）. In the overall group， the ≤ 3 cm group， the 3-5 cm group， the 0-2 cm group， and the 2-3 cm group of the tumor maximum diameter， the modified classification significantly improved the sensitivity ［（92.61%， 92.41%， 93.33%， 90.30%， 95.04%） vs. （62.27%， 61.24%， 65.56%， 58.79%， 64.46%）］， accuracy ［（90.5%， 89.52%， 91.84%， 87.78%， 91.60%） vs. （62.38%， 60.38%， 67.35%， 58.89%， 63.36%）］， and the area under the curve value （AUC） ［（0.766， 0.742， 0.842， 0.752， 0.726） vs. （0.630， 0.586， 0.765， 0.594， 0.572）］. The positive predictive value （PPV） was more than 96%. Conclusions The modified CEUS LI-RADS classification significantly enhanced the sensitivity， accuracy， and AUC of ≤ 5 cm HCC diagnosis without compromising the PPV. The same applies to ≤ 2 cm HCC diagnosis， which would provide more reliable evidence for clinical decision-making.