Non-contrast magnetic resonance imaging for diagnosing shoulder injuries.

PURPOSE To compare preoperative non-contrast magnetic resonance imaging (MRI) with arthroscopy findings in diagnosing labral and rotator cuff tears. METHODS 86 men and 60 women aged 21 to 70 (mean, 52) years underwent non-contrast MRI before arthroscopic operations on the glenohumeral joint. Slices were made in a transverse, parasagittal, and paracoronar orientation. The sequences used were T2- and proton-weighted for paracoronar imaging, T1- and T2-weighted for transverse and parasagittal imaging, and T2-weigthed sequences with fat suppression and short tau inversion recovery sequences. MRI was evaluated with the surgeon to eliminate interobserver bias. Arthroscopic surgery was performed by a single surgeon. If a labral or rotator cuff tear was found, surgery was performed using corkscrew anchors. RESULTS For full thickness rotator cuff tears, MRI and arthroscopy detected them in 76 and 82 patients, respectively. One such tear found by MRI could not be confirmed by arthroscopy. MRI missed 4 subscapularis and 3 supraspinatus tears. The sensitivity and specificity of MRI in diagnosing full thickness rotator cuff tears were 0.90 and 0.91, respectively. For labral tears, MRI and arthroscopy detected them in 16 and 31 patients, respectively. One anterior labral tear detected by MRI could not be verified by arthroscopy. All 16 labral tears detected by MRI were Bankart type-I tears (of the anterior glenoid) except for one superior labral tear from anterior to posterior (SLAP tear). All 13 SLAP tears (10 type 2 and 3 type 3) except for one could be found by arthroscopy only. The sensitivity and specificity of MRI in diagnosing labral tears were 0.52 and 0.89, respectively. CONCLUSION Non-contrast MRI is reliable only for diagnosing full thickness rotator cuff tears and anterior labral tears. Direct or indirect contrast enhancement is recommended for more differentiation. Special scan orientation is necessary for SLAP tears.