Multichannel deep learning MRI model predicts breast cancer axillary node metastasis, comparable to Node-RADS

Multichannel deep learning MRI model predicted axillary lymph node metastasis with AUC 0.908 (95% CI 0.862-0.954) in external validation, comparable to Node-RADS (AUC 0.909, P=0.971).

• The model remained stable across patient subgroups stratified by age, tumor size, and BI-RADS category.
• Agreement between the AI model and Node-RADS was moderate (kappa = 0.523).

Automated summary

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