Clinical Report: Multi-Scale Feature Enhancement Network for Detecting Lower Limb Fractures in X-Ray Imaging
Overview
The proposed MFRNet demonstrates high detection accuracy for lower limb fractures, achieving 89.1% mAP with only 3.8 million parameters.
Background
Lower limb fractures are prevalent in orthopedic emergencies, often resulting from traffic accidents, sports injuries, and falls. Rapid and accurate diagnosis is crucial to prevent complications such as malunion or nonunion, which can severely affect patients' mobility and quality of life. Traditional X-ray imaging, while cost-effective, presents challenges in detecting subtle fractures due to complex anatomy and low contrast.
Data Highlights
Metric
Value
mAP
89.1%
mAP50-95
52.2%
Parameters
3.8 M
Key Findings
MFRNet utilizes an Adaptive Feature Perception Block (AFPB) to enhance fracture features and reduce background noise.
The Multi-Scale Dilated Attention Module (MSDAM) expands the receptive field to capture multi-scale contextual information.
The model balances high detection accuracy with parameter efficiency.
Future work will focus on multi-center validation and mobile deployment of the model.
Clinical Implications
The MFRNet model offers a promising tool for computer-aided diagnosis in detecting lower limb fractures, potentially improving diagnostic accuracy in clinical settings. Its efficiency may facilitate quicker decision-making in emergency situations.
Conclusion
MFRNet represents a significant advancement in the detection of lower limb fractures in X-ray imaging, combining high accuracy with a compact model size. Further validation and deployment are anticipated to enhance its clinical utility.
