Multi-scale feature refinement network for lower limb fracture detection in X-ray images - Takeaways - MDSpire

Multi-scale feature refinement network for lower limb fracture detection in X-ray images

  • By

  • Zhengguo Wan

  • Yanling Wang

  • Rong Tang

  • Ke Zhang

  • Penghua Liu

  • Zheyu Zhao

  • Yujie Shi

  • Huaran Huo

  • July 2, 2026

  • 0 min

Share

  • 1

    Lower limb fractures are prevalent orthopedic emergencies, necessitating rapid and accurate diagnosis for effective clinical management.

  • 2

    The proposed MFRNet utilizes an Adaptive Feature Perception Block and a Multi-Scale Dilated Attention Module to enhance fracture detection.

  • 3

    MFRNet achieves 89.1% mAP and 52.2% mAP50-95 with only 3.8 M parameters, outperforming existing object detection models.

  • 4

    The model addresses challenges like varying fracture scales and ambiguous boundaries, improving feature discriminability.

  • 5

    Future research will focus on multi-center validation, fracture type classification, and mobile deployment of MFRNet.

Original Source(s)

Multi-scale feature refinement network for lower limb fracture detection in X-ray images

  • by Zhengguo Wan, Yanling Wang, Rong Tang, Ke Zhang, Penghua Liu, Zheyu Zhao, Yujie Shi, Huaran Huo

  • July 2, 2026

Related Content

Reducing time from presentation to diagnosis of scaphoid fractures with cone beam CT: a before-and-after study

Multidisciplinary management promoting wound healing in a patient with Proteus syndrome: a case report

Clavicular Gorham-Stout disease: a rare case report and literature review