ABSTRACT
The purpose of this study was to evaluate the ability of 3.0-Tesla magnetic resonance imaging (MRI) to accurately assess knee articular cartilage lesions. Sixteen patients who had knee 3.0-T MRI and underwent knee arthroscopy for partial meniscectomy were included. Three fellowship-trained sports medicine orthopedic surgeons reviewed all images. Articular lesions on MRI were graded from I to IV and compared with arthroscopic grading using the Outerbridge and the International Cartilage Repair Society (ICRS) classifications. The articular surface was divided into 6 regions. Based on MRI findings, of the 288 articular surface evaluations, 113 (39%) surface evaluations were classified as disease-positive (grade 2 to 4). Kappa interrater reliability scores for MRI evaluation, Outerbridge classification, and ICRS classification were 0.13, 0.54, and 0.41, respectively. Using the Outerbridge classification as a reference standard, the sensitivity, specificity, and accuracy were 57%, 71%, and 63%, respectively. Using the ICRS classification, sensitivity, specificity, and accuracy were 59%, 71%, and 69%, respectively. When isolating the articular grading to the senior author on MRI evaluation vs Outerbridge classification, the sensitivity, specificity, and accuracy were 54%, 92%, and 75%, respectively. Based on the current findings, 3.0-T MRI is as an invaluable noninvasive tool with good diagnostic value for assessing articular cartilage lesions of the knee, although it may not be as sensitive and accurate as previously reported.
