Abstract: Objective To investigate the efficiency of liver imaging report data system version 2018(LI-RADSv2108)in the diagnosis of small hepatocellular carcinoma (sHCC) and the consistency between LI-RADSv2108 and radiologists. Methods One hundred and fifteen patients admitted to our hospital from June 2017 to July 2020 were selected, including 89 HCC patients, 9 non-HCC malignant tumors and 17 benign lesions. The images were analyzed retrospectively with blind evaluation by 2 radiologists, and all the lesions were classified according to standard of LI-RADSv2108. The consistency of the conclusion drawn by the 2 radiologists was analyzed by Kappa statistic. Taking the postoperative pathological results as gold standard, the sensitivity, specificity and accuracy of the diagnositic conclusions drawn by the 2 radiologists using LI-RADSv2108 were calculated. McNemar test was performed for the differences. Results The consistency of the diagnositc results between the 2 radiologists was substantial, Kappa value was 0.748 (95% CI: 0.692-0.876). Using LR-5 as a predictor of sHCC, the sensitivity, specificity and accuracy of the diagnositic conclusion drawn by the 2 radiologists respectively using LI-RADSv2108 were 61.8%, 96.2%, 69.6% and 58.4%, 88.5%, 65.2%, respectively, with no significant difference (P>0.05); Using LR 4/5 as a predictor of sHCC, the sensitivity, specificity and accuracy were 78.8%, 80.8%, 79.1% and 78.8%, 65.4%, 75.6%, respectively, with no significant difference (P>0.05). Conclusion LI-RSDSv2018 has moderate sensitivity and high specificity in diagnosing sHCC and the consistency between LI-RADSv2108 and radiologists is good.

Key words: Hepatocellular carcinoma, Liver imaging report data system, Magnetic resonance imaging, X-ray computer tomography