The purpose of this study was to evaluate the accuracy and the usefulness of MRI in patients with shoulder injuries.

The sensitivity and the accuracy of MRI in the diagnosis of various lesion were evaluated in 20 patients with shoulder injuries.

MRI enabled Bankart lesions and SLAP lesions to be accurately predicted with 100% sensitivity and 90% accuracy in both. However, for types of SLAP lesions, the sensitivity and the accuracy were 60% and 70%, respectively, and for full-thickness or partial-thickness tear of rotator cuff, the sensitivity and the accuracy were 75% in both. MRI was also effective at predicting rotator cuff tear with 100% sensitivity and 95% accuracy. Sensitivity and accuracy for the diagnosis of Hill-Sachs lesion were 91.7% and 90%, respectively. For all lesions except Hill-Sachs lesion, the negative prediction value was higher than the positive prediction value, which means MRI overestimated the lesions.

MRI was accurate and useful in the prediction of prediction of rotator cuff tear, labral tear and Hill-Sachs lesions. However, for the identification of full-thickness or partial-thickness tear of rotator cuff, and types of SLAP lesions, it was less accurate and less useful.