Cross-attention guided multi-modal network for breast ultrasound diagnosis incorporating objective clinical semantics - Takeaways - MDSpire

Cross-attention guided multi-modal network for breast ultrasound diagnosis incorporating objective clinical semantics

  • By

  • Qin Sun

  • Xiaoman Wu

  • Ju Chen

  • Yuhang Zhang

  • Chao Zhou

  • Zheng Zhu

  • June 9, 2026

  • 0 min

Share

  • 1

    CGA-Net is a multi-modal framework that integrates visual data with clinical descriptors using a cross-attention mechanism.

  • 2

    The model demonstrated a robust Out-Of-Fold AUC of 0.905 and an overall accuracy of 0.857 in breast ultrasound diagnosis.

  • 3

    CGA-Net outperformed both clinical-only and image-only baselines, achieving a peak overall ranking AUC of 0.915 when pre-trained.

  • 4

    The framework aims to reduce false positive diagnoses by aligning model attention with expert focus on tumor characteristics.

  • 5

    CGA-Net offers a reliable, interpretable tool for clinicians, addressing operator subjectivity in breast ultrasound diagnosis.

Original Source(s)

Cross-attention guided multi-modal network for breast ultrasound diagnosis incorporating objective clinical semantics

  • by Qin Sun, Xiaoman Wu, Ju Chen, Yuhang Zhang, Chao Zhou, Zheng Zhu

  • June 9, 2026

Related Content

First-line durvalumab in combination with trastuzumab deruxtecan in women with locally advanced unresectable or metastatic, hormone-receptor-negative, HER2-low breast cancer: multicenter, open-label, phase 1b/2 BEGONIA platform trial

RSPO2-induced ferroptosis via PTBP1-mediated FSP1 mRNA decay suppresses breast cancer progression

Correction: Accessibility to molecular studies in gliomas: a challenge for precision medicine in low-resource countries