Optimizing diagnostic imaging data using LI-RADS and the Likert scale in patients with hepatocellular carcinoma

Abstract

IntroductionThe study aimed to compare the diagnostic performance of the Liver Imaging Reporting and Data System (LI-RADS), which incorporates fixed criteria, and the Likert scale (LS), which mainly depends on an overall impression in liver lesion diagnosis.Material and methodsDiagnostic data of 110 hepatic nodules in 103 high-risk patients for hepatocellular carcinoma (HCC) were included. Data including diameter, arterial hyperenhancement, washout, and capsule were reviewed by 2 readers using LI-RADS and LS (range, score 1-5). Inter-reader agreement (IRA), intraclass agreement (ICA), and diagnostic performance were determined by Fleiss, Cohen’s , and logistic regression, respectively.ResultsThere were 53 triphasic enhanced computed tomography (CT) and 50 dynamic magnetic resonance (MR) examinations. Overall, IRA was excellent ( = 0.898). IRA was good for arterial hyperenhancement ( = 0.705), washout ( = 0.763), and capsule ( = 0.771) and excellent for diameter ( = 0.981) and tumour embolus ( = 0.927). Overall, ICA between LI-RADS and LS was fair 0.32; ICA was good for scores of 1 ( = 0.682), fair for scores of 2 ( = 0.36), moderate for scores of 5 ( = 0.52), but no agreement was found for scores of 3 ( = –0.059) and 4 ( = –0.022). LIRADS produced relatively high accuracy (87.3% vs. 80%), relatively low sensitivity (84.3% vs. 98%), and significantly higher specificity (89.83% vs. 64.4%) and positive likelihood ratio (+LR: 8.29 vs. 2.75) compared to LS approach.ConclusionsLI-RADS revealed higher diagnostic accuracy as compared to LS with statistical proof higher specificity and +LR showing its ability to foretell malignancy in high-risk patients. We recommend the practical application of the LI-RADS system in the detection and treatment response monitoring of patients with HCC.