Objective To explore the value of clinical-MRI feature nomograms in the differential diagnosis of benign phyllodes tumor (BPT) and pure mucinous breast carcinoma (PMBC).

Methods The clinical and MRI data of 42 cases of BPT and 40 cases of PMBC confirmed by surgery and pathology were retrospectively analyzed. The age, location, menstrual status and multimodal MRI features between two groups were compared. The independent predictors with significant differential value were selected using multivariate logistic regression analysis, and a nomogram model was constructed. Receiver operating characteristic curve was used to evaluate the predictive performance of the model.

Results There were statistically significant in the age, maximum diameter, T2 weighted imaging (T2WI) low signal separation, apparent diffusion coefficient (ADC) value and internal enhancement weredifferent between two groups (P ＜ 0.05). Among them, the maximum diameter, ADC value and internal enhancement were the independent predictors. The area under curve of the nomogram model was 0.896, the sensitivity and specificity were 90.0% and 80.95%, respectively, and the positive predictive value, negative predictive value and accuracy were 81.82%, 89.47% and 85.37%, respectively.

Conclusions The nomogram based on clinical and multimodal MRI signs has a high reference value in differentiating BPT from PMBC.