The Value Of Ultrasound–based Radiomics In The Differential Diagnosis Of Phyllodes Tumor And Fibroadenoma In The Breast

Objective: Discussing the differentiation of breast phyllodes tumor from fibroadenoma by the radiomic features extracted from ultrasound images.Methods: We retrospectively collected 39 patients with breast phyllodes tumor and 105 patients with breast fibroadenoma in our hospital.The patients were randomly divided into training set and test set.A total of 5234 radiomics features were extracted from ultrasound images,using Least absolute shrinkage and selection operator(LASSO)method,the optimal radiomics model was established through logistic regression,and the diagnostic efficiency of the model was evaluated.Results: The best model of ultrasound imaging to distinguish breast phyllodes tumors from fibroadenoma was Spearman0.85+LASSO+Logistics Regression.The AUC of this model in the training set was 0.917.The AUC in the test set was 0.802.The best model of Ultrasound–based Radiomics to distinguish benign phyllodes tumors and fibroadenoma of the breast was Spearman0.85+LASSO+Logistics Regression.The AUC of this model in the training set was 0.911,while the AUC in the test set was 0.75.Conclusion: As a non-invasive method,Ultrasound–based Radiomics shows good diagnostic efficiency in distinguishing breast phyllodes tumor from fibroadenoma.