Purpose We developed a radiomics strategy that incorporating radiomics features extracted from dual-view mammograms and clinical parameters for identifying benign and malignant breast lesions, and validated whether the radiomics assessment can improve the accurate diagnosis of breast cancer.Methods A total of 380 patients with 621 breast lesions utilizing mammograms on craniocaudal (CC) and mediolateral oblique (MLO) views were randomly allocated into the training (n=486) and testing (n=135) sets in this retrospective study. A total of 1184 and 2368 radiomics features were extracted from singleposition ROI and position-paired ROI, respectively. Clinical parameters were then combined for better prediction. The recursive feature elimination (RFE) and least absolute shrinkage and selection operator (LASSO) methods were used to select optimal predictive features. Random forest algorithm was used to conduct the predictive model, and the performance was evaluated with area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, speci city and accuracy.Results After preprocessing, 467 radiomics features and clinical parameters remained in the single-view and dual-view models. The random forest model using a combination of dual-view radiomics and clinical parameters achieved a favorable performance (area under curve [AUC]: 0.804, 95% con dence interval [CI]: 0.668-0.916) in the distinction of benign and malignant breast lesions, which outperformed single-view model and model without clinical parameters.Conclusions Incorporating with radiomics features of dual-view (CC&MLO) mammogram, age, breast density and type of suspicious lesions can provide a non-invasive approach to evaluate the malignancy prediction of breast lesions, which can facilitate clinical decision making.

Purpose We developed a radiomics strategy that incorporating radiomics features extracted from dual-view mammograms and clinical parameters for identifying benign and malignant breast lesions, and validated whether the radiomics assessment can improve the accurate diagnosis of breast cancer. Methods A total of 380 patients with 621 breast lesions utilizing mammograms on craniocaudal (CC) and mediolateral oblique (MLO) views were randomly allocated into the training (n=486) and testing (n=135) sets in this retrospective study. A total of 1184 and 2368 radiomics features were extracted from single-position ROI and position-paired ROI, respectively. Clinical parameters were then combined for better prediction. The recursive feature elimination (RFE) and least absolute shrinkage and selection operator (LASSO) methods were used to select optimal predictive features. Random forest algorithm was used to conduct the predictive model, and the performance was evaluated with area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, specificity and accuracy. Results After preprocessing, 467 radiomics features and clinical parameters remained in the single-view and dual-view models. The random forest model using a combination of dual-view radiomics and clinical parameters achieved a favorable performance (area under curve [AUC]: 0.804, 95% confidence interval [CI]: 0.668-0.916) in the distinction of benign and malignant breast lesions, which outperformed single-view model and model without clinical parameters. Conclusions Incorporating with radiomics features of dual-view (CC&MLO) mammogram, age, breast density and type of suspicious lesions can provide a non-invasive approach to evaluate the malignancy prediction of breast lesions, which can facilitate clinical decision making.