Related Experiment Video
Updated: May 24, 2025
04:19
Minimally Invasive Treatment for Thoracolumbar Burst Fracture Using Sagittal Alignment Screws and A Trauma Reduction Device
Published on: November 8, 2024
339
AI-based biplane X-ray image-guided method for distal radius fracture reduction
Qing Zha1,2, Sizhou Shen3, Ziyang Ma1,2
1School of Biomedical Engineering (Suzhou), Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, China.
Frontiers in Bioengineering and Biotechnology
|March 6, 2025
Summary
An AI method accurately calculates distal radius fracture (DRF) parameters from X-rays, improving diagnosis and aiding fracture reduction robots. This technology enhances diagnostic accuracy for physicians.
Area of Science:
- Orthopedic surgery
- Medical imaging
- Artificial intelligence
Background:
- Manual reduction of distal radius fractures (DRF) often relies on tactile perception, which can lead to misdiagnosis.
- Accurate assessment of fracture severity and reduction success requires image analysis software, but real-time application during procedures is challenging.
- There is a need for AI-based methods to provide real-time calculation and display of fracture parameters, especially for fracture reduction machines.
Purpose of the Study:
- To develop and evaluate an AI-based method for automatically calculating radiographic parameters in distal radius fractures (DRF).
- To compare the utility and accuracy of different neural network structures for image segmentation in DRF analysis.
- To integrate automated parameter calculation into a system for real-time feedback during fracture reduction.
Main Methods:
- Collected and preprocessed anteroposterior (AP) and lateral (LAT) X-ray images of DRF patients.
- Compared UNet, DeeplabV3+, PSPNet, and TransUNet for semantic segmentation of radius and ulna.
- Utilized the UNet model for segmentation, extracted contours with OpenCV, detected key points, calculated principal axes, and computed parameters (RA, RL, UV, PT).
Main Results:
- The UNet model was selected as the core algorithm due to its performance.
- Achieved segmentation accuracy of 91.31% for the radius and 88.63% for the ulna in AP and LAT X-ray images.
- Demonstrated average errors of -1.36° (RA), -1.7 mm (RL), 0.66 mm (UV), and -1.06° (PT) compared to manual annotations.
- The system was initially deployed on a computer controlling a radial fracture repositioning robot.
Conclusions:
- The developed automated parameter calculation method accurately assesses DRF diagnostic parameters.
- This AI method can be integrated into image-guided reduction processes for fracture rehabilitation robots.
- The technology has the potential to become an intelligent diagnostic tool, enhancing diagnostic accuracy for physicians treating distal radius fractures.
