Abstract / Summary
The reliability of automated three-dimensional magnetic resonance imaging (MRI) analysis of meniscal morphology across different systems remains unclear. The present study investigated inter-model variations in meniscal measurements across five 3-Tesla MRI systems. Ten healthy volunteers (age: 22–60 years) underwent knee MRI scans using five different MRI models. Using fully automated three-dimensional analysis software, the width, thickness, and volume of the medial and lateral meniscus were measured in the anterior, middle, and posterior segments. Inter-model differences were evaluated using the Kruskal–Wallis test. Inter-measurement errors, defined as the absolute difference between each value and the median, were evaluated using histograms. Contributing factors to significant differences among models were also analyzed. Meniscal width and volume were consistent across models ( p > 0.05), with a significant difference in thickness ( p < 0.05) in the middle medial and anterior lateral segments. Among 300 measurements, the width error was ≤ 1.0 mm in 91%, the thickness error was ≤ 0.50 mm in 91%, and the volume error was ≤ 150 mm³ in 97%. Some variations in thickness were attributable to inherent limitations in the resolution of MR images and segmentation errors. These findings suggest that three-dimensional MRI analysis of meniscal morphology provides largely consistent results across different systems, although model-dependent variation was observed in selected thickness measurements. This robust reproducibility may facilitate future investigations of meniscal morphology in clinical and multicenter MRI studies.