Abstract: Objective To explore the clinical value of magnetic resonance imaging (MRI) in the diagnosis of rotator cuff injury. 
Methods One hundred and twenty patients with suspected rotator cuff injury were selected, and all of them were given MRI examination, to analyze the value of MRI in the diagnosis of rotator cuff injury. In the meantime, patients with confirmed rotator cuff injury were given non-surgical treatment, and the difference of clinical imaging data between patients with effective and ineffective treatment were compared. 
Results The sensitivity, specificity, accuracy, positive predictive value and negative predictive value of MRI in diagnosing rotator cuff injury were 98.04%, 94.44%, 97.50%, 99.01% and 89.47%, respectively. The detection rates of infraspinatus tendon injury and subscapular tendon injury in oblique coronal scan were 52.94% and 49.02%, respectively, which were significantly lower than those in axial and oblique sagittal scan (P＜0.05). The detection rate of supraspinatus tendon injury in oblique sagittal scan was 77.45%, which was significantly lower than that in oblique coronal scan (P＜0.05). The detection rates of teres minor tendon injury in axial and oblique coronal scan were 7.84% and 13.73%, respectively, which was significantly lower than those in oblique sagittal scan (P＜0.05). The age of patients with ineffective treatment was (65.54±7.03) years, and the retraction distance of supraspinatus muscle was (24.41±5.93) mm, which were significantly higher than those in patients with effective treatment (P＜0.05). The degree of subscapular muscle tear in patients with ineffective treatment was significantly greater than that in patients with effective treatment (P＜0.05), and the complete tear ratio was 80.95%. Logistic regression analysis showed that age, and degree of subscapular muscle tear were the influencing factors of non-surgical treatment effect in patients with rotator cuff injury (P＜0.05). Logistic regression model was constructed, and the area under the ROC curve of the model in predicting ineffective treatment was 0.813 (95%CI: 0.728-0.898, P＜0.05), with sensitivity and specificity of 88.00% and 64.50%, respectively. 
Conclusion MRI has high sensitivity and specificity in the diagnosis of rotator cuff injury, and the MRI parameter, subscapular muscle tear, is related to the effect of non-surgical treatment, which is worthy of further study. 

Key words: rotator cuff injuries, magnetic resonance imaging, diagnosis