ABSTRACT
The objective of this study was to compare the diagnostic performance of a dedicated orthopedic magnetic resonance (MR) imaging system (0.18 T) and a conventional MR imaging system (1.0 T) in the detection of articular cartilage lesions. Fifty knee joint specimens of pigs with artificially created articular cartilage lesions of different diameters, grades (2-3), and localizations, as well as 50 joints with intact articular cartilage, were imaged at 0. 18 and 1.0 T. Diagnostic performance was determined by means of receiver operating characteristics (ROC) analysis with three independent observers. For none of the pulse sequences used at 0.18 T or 1.0 T areas under ROC curves (A(z)) showed significant differences between the three observers. A(z) values from averaged data were as follows: a) 0.18 T: T1-weighted spin echo (SE): 0.70, proton-density-weighted SE: 0.59, T2-weighted SE: 0.61, two-dimensional (2D) gradient-echo (GRE): 0.73, 3D GRE: 0.75; and b) 1.0 T: T1-weighted SE: 0.73, fat-suppressed T2-weighted turbo-SE: 0. 79, 2D fast low-angle shot (FLASH): 0.79, fat-suppressed 3D FLASH: 0. 96, and water-excited 3D double-echo steady state (DESS): 0.96. With the use of 3D pulse sequences, the high-field system demonstrated a significantly better diagnostic performance than the low-field system in the detection of grades 2 and 3 articular cartilage lesions (P < 0.001).
