Multimodal deep learning with few-shot framework predicts HER2-low status in triple-negative breast cancer

Multimodal MRI/US fusion model achieved AUC 0.877 for classifying HER2-low vs HER2-zero in TNBC; hybrid model adding postop clinical variable reached AUC 0.884 in this small retrospective development study (n=122).

• Retrospective study of 122 patients with triple-negative breast cancer, utilizing a few-shot learning framework (FADNET).
• Pre-treatment imaging-only fusion (US + multiparametric MRI) AUC 0.877; single-modality AUCs ranged 0.764–0.806.
• Major limitation: single-center, small sample, no external validation; not yet clinically deployable.

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