Purpose: To evaluate and compare the diagnostic accuracy of appropriate magnetic resonance (MR) sequences in the detection of cartilage lesions at 1.5T and 3.0T.
Material and methods: Twelve chondral defects of varying depths, widths, and locations were created in the retropatellar hyaline cartilage in six sheep cadaver limbs. Axial images employing three fat-suppressed imaging sequences--(1) a T2-weighted fast spin-echo (FSE) sequence, (2) a two-dimensional (2D) and (3) three-dimensional (3D) gradient-echo (GE) sequence at 1.5T and 3.0T using an extremity quadrature coil--were evaluated by three experienced radiologists. Statistical analysis of the results consisted of receiver operating characteristics (ROC) and significant testing using the bivariate chi-square test. In addition, signal-to-noise ratios (SNR) and contrast-to-noise ratios (CNR) were evaluated with significance testing using the Wilcoxon test.
Results: The 3D GE sequence compared favorably with other sequences at 3.0T and 1.5T (Az=0.88 at 3.0T and Az=0.85 at 1.5T) missing only one small grade 2 lesion. 2D GE imaging was inferior to 3D imaging at both field strengths (P<0.05) in general. However, compared to 1.5T, lesion detectability was improved at the higher magnetic field of 3.0T (Az=0.81 and 0.73 at 3.0T and 1.5T, respectively). FSE images showed significantly inferior sensitivity and less anatomical detail compared to the GE sequences at both field strengths (Az=0.64 and 0.72 at 3.0T and 1.5T, respectively; P<0.05). However, compared to 1.5T, lesion detectability SNR and CNR values were superior in all sequences tested at 3.0T.
Conclusion: MRI at 3.0T improves SNR and CNR significantly in the most common sequences for cartilage MRI, resulting in an improvement in chondral lesion detection. GE imaging therefore allows resolution to be increased in an acceptable time manner for patient comfort, and the 3D GE fat-suppressed sequence at 3.0T appears to be best suited for cartilage imaging in a clinical setting.