Objectives: The aim of this study was to evaluate the diagnostic performance of introducing diffusion-weighted imaging (DWI) as a major feature to extracellular agent (ECA)-MRI for diagnosing HCC in comparison with gadoxetic acid (hepatobiliary agent, HBA)-MRI using Liver Imaging Reporting and Data System (LI-RADS) v2018. Methods: This was a prospective intra-individual comparison study using two different types of contrast agents for liver MRI conducted at a tertiary referral academic center. One hundred forty-seven observations in 122 patients at high risk for HCC scheduled for liver surgery were included. The sensitivity, specificity, and accuracy of LI-RADS category 5 (LR-5) for HCC diagnosis according to conventional and modified LI-RADS on ECA- and HBA-MRI were measured and compared. Modified LI-RADS incorporated hyperintensity on DWI as a major feature with ECA-MRI, and hypointensity on transitional phase (TP) and/or hepatobiliary phase (HBP) as washout appearance on HBA-MRI, respectively. Results: Modified LI-RADS on ECA-MRI had higher sensitivity and accuracy than modified LI-RADS on HBA-MRI (90.3% vs. 74.9%, p < 0.001; and 91.9% vs. 76.9%, p < 0.001, respectively), as well as higher specificity, although the difference did not reach statistical significance (96.0% vs. 88.0%, p = 0.157). The specificity of modified LI-RADS ECA-MRI was slightly lower than both conventional criteria but without a significant difference (96.0% vs. 100%, p = 0.317). Conclusions: Using DWI findings as a major feature for modified LR-5 on ECA-MRI showed better sensitivity and accuracy than modified LR-5 on HBA-MRI, without significantly compromising specificity compared with conventional LR-5 on ECA- or HBA-MRI. Key Points: • Prospective intra-individual comparison study using two different types of contrast agents, extracellular agent (ECA) and gadoxetic acid (hepatobiliary agent, HBA), for liver MRI was conducted. • Applying diffusion restriction of a hepatic observation on ECA-MRI as a major feature of LI-RADS v2018 resulted in higher sensitivity and accuracy of LR-5 observations for HCC diagnosis than conventional LI-RADS v2018, and even compared to modified LI-RADS using modified washout on HBA-MRI. • Despite increase in sensitivity and accuracy of LR-5 observations on modified LI-RADS on ECA-MRI, the specificity was not significantly different compared with conventional LI-RADS.
