MR diagnosis of meniscal tears of the knee: value of axial three-dimensional Fourier transformation GRASS images.

Axial MR images of 40 knees in 37 patients with clinically suspected meniscal tears were obtained by using a three-dimensional (3-D) Fourier transform, gradient-refocused acquisition in the steady state (GRASS) pulse sequence. All knees also were examined in both coronal and sagittal planes with a two-dimensional (2-D) Fourier transform, spin-echo pulse sequence for T1-weighted images and a multiplanar GRASS pulse sequence for T2*-weighted images (conventional 2-D pulse sequences). Arthroscopic confirmation was available in all cases and was used as the gold standard. In these 40 knees, 32 of 33 meniscal tears and 45 of 47 normal menisci were correctly identified with axial 3-D imaging, yielding a sensitivity of 97% and specificity of 96%. Conversely, the sensitivity was 82% and the specificity was 100% for the 2-D technique in the diagnosis of meniscal tears. A combination of both techniques yielded 100% sensitivity, 100% specificity, and 100% accuracy in the evaluation of the menisci. Seven discrepancies between the findings of the two techniques were found in the evaluation of 33 arthroscopically proved torn menisci; six of the seven represented false-negative 2-D images and one was a false-negative axial 3-D image. Of 47 normal menisci, only two false-positives occurred with axial 3-D imaging and none with 2-D imaging. Despite the lack of a statistically significant difference, these preliminary findings suggest that the sensitivity in the diagnosis of meniscal tears can be improved by adding thin-sliced axial images with the 3-D GRASS pulse sequence to conventional 2-D imaging.