Value Of Radiomics Based On DCE-MRI In Predicting Luminal And Non-luminal Breast Cancer

Objective: To investigate the value of radiomics based on dynamic contrast enhanced magnetic resonance imaging(DCE-MRI)in predicting Luminal and non-Luminal breast cancer.Methods: We retrospectively analyzed images of breast DCE-MRI of 42 patients with breast cancer confirmed by pathology.The strongest enhancement phase was selected according to the time-signal intensity curve(TIC)and was imported into RadCloud software for analysis.The lesions were outlined by manually segmentation,and radiomic features were extracted.All cases were divided into Luminal group and non-Luminal group according to the expression of estrogen receptor(ER)and progesterone receptor(PR).Then the least absolute shrinkage and selection operator(LASSO)was used to select the most important features and construct the radiomics label.Spearman correlation analysis was used to evaluate the correlation between the screened characteristic values with ER,PR and human epidermal growth factor receptor-2(HER2).Classifier were constructed based on logistic regression(LR),random forest(RF),K nearest neighbor(KNN)and support vector machine(SVM),and their prediction performance were verified by 5-fold cross validation,and the sensitivity,specificity,accuracy and area under the curve(AUC)of the classifier were calculated.Results: Among the 42 patients,32 were divided into Luminal and 10 were divided into non-Luminal.4 optimal features,including 1 shape feature and 3 features based wavelet,were statistically significant in separating Luminal from non-Luminal cancers,and 3 features,including the Original_Shape_Maximum 2D Diameter Row,Wavelet-LLH_GLCM_Idn and Wavelet-HHH_GLCM_Correlation had significant correlations with ER and PR levels,while the 4 features had no significant correlations with HER2 levels.The classifier based on RF has the best prediction performance for Luminal breast cancer.Its sensitivity,specificity,accuracy and AUC were 0.838,0.900,0.853 and 0.876,respectively.Conclusion: Radiomics based on DCE-MRI can predict Luminal and non-Luminal breast cancer,and the prediction performance of the prediction model based on RF is the best.