Related Experiment Video
Updated: Aug 6, 2026
06:45
Automated Joint Space Detection Improves Bone Segmentation Accuracy
Published on: November 28, 2025
Multi-scale feature refinement network for lower limb fracture detection in X-ray images
Zhengguo Wan1, Yanling Wang1, Rong Tang2
1CT and MRI Department, Handan First Hospital, Handan, China.
Frontiers in Medicine
|July 17, 2026
Summary
A new network, MFRNet, improves lower limb fracture detection by refining multi-scale features. This computer-aided diagnosis tool offers high accuracy and efficiency for orthopedic emergencies.
Area of Science:
- Orthopedic imaging and computer-aided diagnosis.
Background:
- Accurate diagnosis of lower limb fractures is crucial but challenging due to variations in fracture scale and ambiguous boundaries.
- Current object detection models struggle with precise fracture detection in X-ray images.
Purpose of the Study:
- To develop an advanced deep learning model for accurate and efficient detection of lower limb fractures.
- To address the limitations of existing methods in handling scale variations and boundary ambiguities.
Main Methods:
- Proposed a novel Multi-scale Feature Refinement Network (MFRNet).
- Introduced the Adaptive Feature Perception Block (AFPB) for enhanced feature extraction and noise suppression.
- Implemented the Multi-Scale Dilated Attention Module (MSDAM) to capture multi-scale contextual information.
Main Results:
- MFRNet achieved 89.1% mAP and 52.2% mAP50-95 on a lower limb fracture X-ray dataset.
- The model demonstrated superior performance compared to mainstream object detection models.
- MFRNet has only 3.8 million parameters, indicating high parameter efficiency.
Conclusions:
- MFRNet offers a promising balance of high detection accuracy and parameter efficiency for lower limb fracture diagnosis.
- The model holds significant clinical practical value for computer-aided diagnosis systems.
- Future research will involve multi-center validation, fracture classification, and mobile deployment.
