To prospectively evaluate the diagnostic accuracy of magnetic resonance (MR) arthrography for the detection of articular cartilage abnormalities at 3.0T and 7.0T in cadaveric wrists. MR imaging (MRI) was performed in nine cadaveric wrists (four right wrists, five left; mean age, 81.0 ± 9.8 years) after the intraarticular administration of gadoterate-meglumine. A 3.0T and 7.0T MR system, mechanically identical custom-built 8-channel wrist coil arrays and a similar standard MRI protocol, were used. MR images were evaluated for visibility of articular cartilage surfaces, presence of cartilage lesions, and confidence of diagnosis by two independent radiologists. Open pathologic inspection served as reference standard. Sensitivity, specificity, negative predictive values (NPV) and positive predictive values (PPV), and accuracy (ACC) were calculated. Wilcoxon signed rank test was used to assess differences in the diagnostic performance. Visibility of articular cartilage surfaces was significantly better at 3.0T than at 7.0T (P < 0.001). Mean sensitivity, specificity, NPV, PPV, ACC for both readers were 63%, 90%, 85%, 76%, 82% at 3.0T, respectively, and 52%, 91%, 82%, 75%, 79% at 7.0T. The difference between 3.0T and 7.0T was not significant for reader 1 (P = 0.51), but was significant for reader 2 (P = 0.01). The level of confidence was significantly higher at 3.0T than at 7.0T for both readers (P = 0.004; P = 0.03). MR arthrography of the wrist at 7.0T is still limited by the lack of commercially available radiofrequency coils and limited experience in sequence optimization, resulting in a significantly lower visibility of anatomy, lower diagnostic accuracy, and level of confidence in judging cartilage lesions compared to 3.0T.
Read full abstract