Thickness of patellofemoral articular cartilage as measured on MR imaging: sequence comparison of accuracy, reproducibility, and interobserver variation

Since the thickness of cartilage is an important indicator of the status, progression and response to therapy of articular disorders, assessment of it is desirable. This study was undertaken to assess the accuracy, precision, and reliability of magnetic resonance (MR) measurements of articular cartilage. Fifteen cadaveric patellas were imaged in the axial plane at 1.5 T. Gradient echo and fat-suppressed FSE, T2-weighted, proton density, and T1-weighted sequences were performed. We measured each 5-mm section separately at three standardized positions, giving a total of 900 measurements. These findings were correlated with independently performed measurements of the corresponding anatomic sections. A hundred random measurements were also evaluated for reproducibility and interobserver variation. Although all sequences were highly accurate (range r = 0.78-0.82), the T1-weighted images were the most accurate, with a mean difference of 0.25 mm and a correlation coefficient of 0.85. All sequences were also highly reproducible (mean difference between -0.09 and 0.05 mm) with little inter-observer variation (mean difference -0.04 and 0.11 mm). In an attempt to improve the accuracy of the MR measurements further, we retrospectively evaluated all measurements with discrepancies greater than 1 mm from the specimen. All these differences were attributable to focal defects causing exaggeration of the thickness on MR imaging. MR imaging is accurate, precise, and reliable as a basis for measuring articular cartilage and may potentially be usable to monitor progression of articular disorders. Care must be taken not to overestimate the thickness of areas with surface defects.