Study on predicting HER2 status in breast cancer using radiomics and deep learning with multiparametric MRI
WANG Wenjiang
LIU Yanjun
WANG Lei
LI Jiaojiao
PANG Zhiying
YANG Fei
CUI Shujun
Abstract：Objective To develop a multimodal model combining radiomics and deep learning with multiparametric MRI for predicting human epidermal growth factor receptor 2 status in breast cancer patients preoperatively.Methods A total of 176 pa-tients with invasive breast cancer who underwent breast MRI and had HER2 status results were included.The dataset was ran-domly split into a training set and a test set at a 7∶3 ratio.Radiomics and deep learning features from T2-weighted imaging(T2WI)and dynamic contrast-enhanced MRI(DCE-MRI)were integrated.The following models were developed using ResNet 50 and Visual Transformer:R_T2,R_DCE,R_mp-MRI,V_T2,V_DCE,and V_mp-MRI.The models were trained using five-fold cross-validation on the training set,and their performance on the test set was evaluated using the area under the receiver op-erating characteristic curve(AUC),accuracy,sensitivity,specificity,positive predictive value,negative predictive value,and F1 score.The DeLong test was used to compare statistical differences within the ResNet 50 and Visual Transformer groups.Results The six models established in this study performed well,with AUC values of 0.70,0.79,0.91,0.78,0.85,and 0.96 for the R_T2,R_DCE,R_mp-MRI,V_T2,V_DCE,and V_mp-MRI models,respectively.The DeLong test showed that the P-values for the comparisons within the ResNet 50 and Visual Transformer groups were all<0.05,indicating statistically signifi-cant differences.Conclusion The multimodal model combining radiomics and deep learning with multiparametric MRI can ef-fectively predict the preoperative HER2 status in breast cancer.
Keywords：Breast cancerHuman epidermal growth factor receptor 2Magnetic resonance imagingDeep learningra-diomics
Publication Date：2025-04-30
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 54-59 )