Inter-Reader Agreement for CT and MRI LI-RADS Version 2018: An Individual Participant Data Meta-Analysis

Area of Science:

• Radiology and Medical Imaging
• Hepatocellular Carcinoma (HCC) Diagnosis
• Data Science in Medicine

Background:

• The Liver Imaging Reporting and Data System (LI-RADS) is crucial for noninvasive hepatocellular carcinoma (HCC) diagnosis.
• Previous single-center studies indicated variability in LI-RADS interpretation, necessitating broader evaluation.

Purpose of the Study:

• To assess inter-reader agreement for LI-RADS CT/MRI v2018 categorization and imaging features.
• Utilize an individual participant data (IPD) meta-analysis approach for both CT and MRI separately.

Main Methods:

• Retrospective meta-analysis of five studies involving 975 observations from the LI-RADS Living IPD Meta-Analysis Project.
• Two-step meta-analytic approach to calculate pooled kappa (κ) for agreement, with heterogeneity assessed using τ² and I².
• Meta-regression analysis explored factors influencing agreement, including reader experience and study country.

Main Results:

• Inter-reader agreement (κ) for major features ranged from 0.21-0.74 (MRI) and 0.11-0.75 (CT).
• Agreement for ancillary features and LR-M varied between -0.07 and 0.68 across modalities.
• LI-RADS categorization agreement was higher for MRI (κ=0.76) compared to CT (κ=0.50).

Conclusions:

• MRI demonstrated consistently higher inter-reader agreement for shared imaging features than CT.
• Overall inter-reader agreement for LI-RADS interpretation remains variable, highlighting areas for improvement in standardization.