Multiparametric MRI radiomics plus deep transfer learning predicts HER2 status preoperatively

Combining intratumoral/peritumoral radiomics and deep transfer learning from multiparametric MRI predicted HER2 status in breast cancer: AUC 0.965 training, 0.904 internal validation, 0.844 and 0.846 external test sets.

• Multicenter study included 411 patients from three centers, with DCE-MRI, T2WI, and DWI.
• The combined model integrating radiomics, deep learning features, and clinical indicators showed best predictive accuracy.
• Visual interpretability was assessed using Grad-CAM and SHAP methods.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.