Abstract:
ObjectiveTo evaluate the differential diagnosis of benign and malignant breast BI-RADS category 3 and above nodules by ultrasound radiomics.
MethodsA total of 164 breast nodules of BI-RADS category 3 and above diagnosed by ultrasound were retrospectively analyzed.The lesions were divided into training set (n=123) and test set (n=41).Base on feature extraction engineering of ultrasound radiomics, the high latitude feature parameters of the region of interest in the image were extracted, and the feature dimension reduction was used to screen the reliable features which established the model for classification research.The training set was used to establish the model, and the test set was used to verify the accuracy of the model.
ResultsThe sensitivity, specificity, accuracy and area under ROC curve of the model established by ultrasound radiomics were 87.51%, 84.62%, 85.00% and 0.854(0.707-0.946), respectively.
ConclusionsUltrasound radiomics has certain diagnostic value in differentiating benign and malignant breast BI-RADS category 3 and above nodules.