Purpose: To conduct a multi-center assessment of the use of a 0.2-T, extremity MR system (E-scan; General Electric Lunar Corp. and Esaote, Genoa, Italy) for identifying tears of the rotator cuff and glenoid labrum.
Materials and methods: A retrospective study was performed involving 160 patients (age range, 15-84 years old) from five facilities in the United States, comparing shoulder MR imaging to surgical findings. MR imaging of the shoulder was conducted as follows: shoulder coil; T1-weighted spin echo, coronal-oblique and axial images; short Tau inversion recovery (STIR), coronal-oblique images and axial images; and T2-weighted spin echo, coronal-oblique and sagittal-oblique images. The MR examinations were interpreted in an independent, prospective manner by two radiologists with extensive musculoskeletal MRI experience. Arthroscopic (N = 103) or open surgical (N = 57) procedures were performed within a mean of 53 days after MR imaging.
Results: Surgical findings demonstrated rotator cuff tears in 131 patients and labral tears in 60 patients. For the rotator cuff, the sensitivity, specificity, positive predictive value, and negative predictive value were 90%, 93%, 98%, and 68%, respectively. For the labrum, the sensitivity, specificity, positive predictive value, and negative predictive value were 55%, 100%, 100%, and 82%, respectively.
Conclusions: There was good agreement when MR results obtained using the extremity MR system were compared to surgical findings for identifying rotator cuff tears, while the sensitivity of MR imaging for determining labral tears was relatively poor. Nevertheless, these findings were comparable to those reported in the peer-reviewed literature for MR systems operating at mid-, and high-field-strengths.
Copyright 2004 Wiley-Liss, Inc.