Objective To explore the correlation between CT features and pathological types, risk classification of thymus epithelial tumors(TETs).

Methods The preoperative clinical and imaging data of patients with TETs proved by pathology were retrospectively analyzed, and the correlation between general data and CT features of TETs and pathological classification, risk classification were compared.Binary logisitic regression was used to analyze the independent risk factors of high-risk TETs, and a nomogram was drawn to predict the high risk of TETs patients.ROC curve was used to verify the accuracy of the nomogram model.

Results The differences of the age, chest pain, peripheral margin, calcification, separation, enhancement amplitude, pericardium thickening, mediastinal lymph node enlargement, surrounding fat infiltration and surrounding mediastinal structure among different pathological types were statistically significnat(P < 0.05).The differences of the chest pain, morphology, outer margin, density, necrosis/cystic change, calcification, arterial CT value, venous CT value, enhancement amplitude, pericardial thickening, pleural thickening, mediastinal lymph node metastasis, peripheral fat infiltration, encircling mediastinal structure and invasion of peripheral blood vessels between the low and high risk groups were were statistically significnat(P < 0.05).The chest pain, outer margin and pericardial thickening were the independent risk factors of high risk TETs, and the AUC of the ROC curve for nomogram model was 0.818(95%CI: 0.712-0.897, P < 0.01).

Conclusions Combined with clinical and CT features, the pathological classification of TETs can be predicted to a certain extent, and the nomogram model constructed by chest pain, peripheral margin and pericardial thickening has better predictive efficacy in patients with high risk of TETs.