Abstract:Objective To evaluate the diagnostic value of enhanced MRI in myomerial infiltration,cervical infiltration and lymph node metastasis of endometrial carcinoma and analyze the misdiagnostic factors.Methods From March 2009 to March 2013,167 patients with endometrial carcinoma in Department of Gynecology of Tianjin Medical University General Hospital were studied retrospectively. All patients underwent enhanced MRI scan before operation, MRI staging were compared with surgicopathologic staging and several factors that make accurate assessment of myometrial infiltration, cervical infiltration and lymph node metastasis difficult by MRI were analyzed. Results (1) The accuracy of MRI varied with the stage of tumor, the advanced stage the tumor, the lower the accuracy. The accuracy decreased as tumor grade increased. The differences of accuracy rate had statistically significant (P<0.05); The accuracy for the endometrioid type and non-endometrioid type were 79.74% and 64.29%, the differences of accuracy rate had statistically significant(P<0.05).(2) The sensitivity, specificity, diagnostic accuracy, positive and negative predictive values and ? value of MRI in identifying superficial myometrial infiltration were 91.79%, 90.91%, 91.62%, 97.62%, 73.17% and 0.758, respectively.Those for deep myometrial infiltration were 90.91%, 91.79%, 91.62%, 73.17%, 97.626% and 0.758, respectively. Those for cervical infiltration were 84.21%, 95.95%,9 4.61%, 72.73%, 97.93% and 0.750,respectively.Those for lymph node metastasis were 45.00%, 91.16%, 85.63%, 40.91%, 92.41% and 0.347,respectively. (3) The main causes of error in enhanced-MRI were myomas, cornual lesions, deep myometrial invasion, large tumor size, non-endometrioid tumor type and lower tumor grade( All P<0.05). Conclusion Enhanced MRI has a high accuracy and a low tendency to produce false negatives in identifying patients with deep myometrial invasion, cervical invasion, and pelvic lymph node metastasis. The main causes of error in preoperative prediction were myomas, cornual lesions, large tumor size, non-endometrioid tumor type, and lower tumor grade.