Objective To evaluate the value of a color Doppler ultrasound-based radiomics model in identifying the malignancy of breast masses.Methods Patients'data were collected from routine breast ultrasound examinations conducted in our department from July 2022 to December 2023.A total of 60 cases with breast masses were included and enrolled into a training set(30 cases)and a validation set(30 cases).Medical imaging 2D/3D visualization using ITK-SANP software was employed to delineate the region of interest(ROI)along the contour of the breast mass on grayscale ultrasound images.These ROIs were then imported into A.K.software for radiomic feature extraction.The least absolute shrinkage and selection operator(LASSO)algorithm was applied for multivariate regression to establish a predictive model.Receiver operating characteristic(ROC)curves were plotted,and the area under the curve(AUC),accuracy(ACC),sensitivity(SEN),and specificity(SPE)were calculated for model evaluation.Results Univariate and multivariate logistic regression analyses revealed that morphology,margins,and microcalcifications were independent factors for predicting the malignancy of breast tumors.Furthermore,based on the LASSO analysis algorithm,the AUC was found to be 0.871 in the training set and 0.849 in the validation set.In the training set,the ACC was 0.728,SEN was 0.961,and SPE was 0.637.In the validation set,the ACC was 0.753,SEN was 0.974,and SPE was 0.607.Conclusion The LASSO model established based on color Doppler ultrasound radiomics demonstrates potential application in the clinical identification of the malignancy of breast masses.
维普
万方数据