Fat-suppressed fast spin-echo mid-TE (TE[effective]=34) MR images: comparison with fast spin-echo T2-weighted images for the diagnosis of tears and anatomic variants of the glenoid labrum

Objective. To compare the sensitivity, specificity, and accuracy of fat-suppressed fast spin-echo (FSE) mid-TE (TE[effective]=34) images with fat-suppressed FSE T2-weighted images for the diagnosis of labral abnormalities. Design and patients. The study included 27 consecutive patients who had axial fat-suppressed FSE T2-weighted and fat-suppressed FSE mid-TE MR images, and had labral abnormalities diagnosed at arthroscopy. The acquisition time was about 5 min for each sequence, but the mid-TE sequence allowed a higher spatial resolution than the T2-weighted images (256×256 versus 256×192). Twenty-eight age-matched patients with arthroscopically normal labra were included as a control group. The labrum was graded on the MR images as normal or abnormal separately by two musculoskeletal radiologists who were masked to the history and arthroscopic results. The surgical findings were used as the gold standard for calculating the sensitivity, specificity, and accuracy for interpreting the correct location of a labral abnormality. The sensitivity, specificity, and accuracy for the two sequences were compared with a McNemar test, and significance defined as P<0.05. Results. For observer 1, the sensitivity for labral abnormalities was 0.59 on the T2-weighted images, and 0.78 on the mid-TE images (P=0.12). The specificity was 0.54 for the T2-weighted, and 0.64 for the mid-TE images (P=0.51). The accuracy was 0.56 for the T2-weighted, and 0.71 for the mid-TE images (P=0.08). For observer 2, the sensitivity/specificity/accuracy was 0.67/0.93/0.80 for the T2-weighted, and 0.70/0.86/0.78 for the mid-TE images (all P>0.5). Conclusion. In this small study there is no statistically significant difference for demonstrating labral abnormalities between FSE T2-weighted images, and higher-resolution fat-suppressed FSE mid-TE (TE[effective]=34) images obtained with a similar acquisition time. Although there was a general trend toward higher sensitivity and accuracy with the mid-TE sequence, particularly for one of the two observers, a larger study is needed to determine whether this is the preferred single axial pulse sequence for conventional MR imaging of the labrum. Received: 28 June 1999 Revision requested: 9 September 1999 Revision received: 20 September 1999 Accepted: 28 September 1999