Grading articular cartilage of the knee using fast spin-echo proton density-weighted MR imaging without fat suppression.

Abstract

The objective of this work was to determine the accuracy of fast spin-echo proton density-weighted MR imaging in the evaluation of the articular cartilage of the knee using arthroscopy as a gold standard. We retrospectively reviewed MR images of the knee in 54 patients for whom arthroscopic results were available. All MR imaging studies included fast spin-echo proton density-weighted coronal and axial sequences as part of our routine protocol. Evaluation of the articular surfaces was performed by three independent observers who were unaware of the arthroscopic results. The cartilage surfaces were graded using a 3-point system, and results were compared with arthroscopic findings. Of 324 cartilage surfaces evaluated, arthroscopy showed 241 surfaces as normal, 56 as containing partial-thickness defects, and 27 as containing full-thickness defects. Compared with arthroscopic data, sensitivity of MR imaging for the three reviewers was 59-73.5%; specificity, 86.7-90.5%; positive predictive value, 60.5-72.6%; negative predictive value, 86.0-90.8%; and accuracy, 79.6-86.1%. Interobserver variability for the presence of disease, which was measured using the kappa statistic, was 0.63. Fast spin-echo proton density-weighted MR imaging sequences can be used to evaluate the cartilage of the knee with accuracy comparable to that of previously reported cartilage-specific sequences.