Abstract: Multivariate Logistic regression analysis was used to construct an MRI multiparametric diagnostic prediction model of breast nodules and to verify the diagnostic efficacy of this model. A total of 205 lesions were retrospective analysis in patients with breast lesions, including 113 malignant lesions and 92 benign lesions. The morphological and kinetic characteristics of the benign and malignant lesions were observed. Using the malignant group as the experimental group and the benign group as the control group, 205 lesions were randomly divided into 174 samples in the training set and 31 samples in the external validation test set. Through statistical analysis, build the prediction model with the training set data and draw its nomogram; use the external verification test set sample to verify the consistency of model diagnosis; draw the receiver operator characteristic (ROC) curve and verify the differentiation of the model by calculating the area under the curve (AUC) to evaluate the sensitivity, specificity and accuracy of the model in the differential diagnosis of benign and malignant breast nodules. After screening, the independent risk factors included in the prediction model were patient age and lesion ADC value, TIC curve, lesion size, and enhancement characteristics. After the study, the prediction accuracy of the prediction model for benign and malignant breast nodules reached 92.2% in this sample. Therefore, the nomogram based on Logistic regression analysis has a high reference value for the prediction of benign and malignant breast nodules.

Key words: breast cancer, dynamic enhancement of the MRI, nomogram, prediction model