Related Experiment Video
Updated: Jul 4, 2025
09:30
Quantitative Magnetic Resonance Imaging of Skeletal Muscle Disease
Published on: December 18, 2016
19.6K
Unsupervised motion artifact correction of turbo spin-echo MRI using deep image prior
Jongyeon Lee1,2, Hyunseok Seo2, Wonil Lee3,4
1School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea.
Magnetic Resonance in Medicine
|January 29, 2024
Summary
This study introduces an unsupervised deep learning method for correcting motion artifacts in MRI scans. The novel approach significantly enhances image quality without requiring large training datasets, demonstrating clinical potential.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Magnetic Resonance Imaging
Background:
- Motion artifacts degrade Magnetic Resonance Imaging (MRI) quality.
- Deep learning correction methods often need extensive, resource-intensive training.
Purpose of the Study:
- To propose an unsupervised deep learning method for motion artifact correction in turbo-spin echo MRI.
- To utilize the deep image prior framework for efficient artifact removal.
Main Methods:
- The method leverages neural network parameterization for motion artifact suppression.
- It involves image parameterization, spatial transformation, and a motion simulation model.
- An optimization process synthesizes corrupted images to match acquired ones, minimizing discrepancies.
Main Results:
- Significant improvements in structural similarity index observed in simulation studies.
- Ablation studies confirmed the effectiveness of individual components in artifact correction.
- Real-world data experiments indicated strong clinical potential.
Conclusions:
- The proposed method effectively corrects rigid and in-plane motion artifacts in turbo spin-echo MRI.
- It offers substantial quantitative and qualitative improvements in image quality.
