Related Experiment Video
Updated: May 23, 2025

Author Spotlight: Using Hyperpolarized Xenon-129 MRI to Study Lung Diseases
Published on: January 5, 2024
Denoising very low-field magnetic resonance images using native noise modeling
Tonny Ssentamu1,2, Alvin Kimbowa3,4, Ronald Omoding4
1Department of Physiology, Makerere University, Kampala, Uganda.
Native Noise Denoising (NND) enhances low-field MRI by using the scanner's own noise to improve image quality. This method boosts signal-to-noise ratio and preserves details, making MRI more accessible in low-resource settings.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Artificial Intelligence in Healthcare
Background:
- Low-field MRI offers cost-effective, portable imaging solutions, crucial for resource-limited healthcare.
- Significant image noise in low-field MRI hinders clinical diagnostic confidence and utility.
- Existing denoising methods may not fully address the unique noise characteristics of low-field systems.
Purpose of the Study:
- To introduce and validate Native Noise Denoising (NND) for enhancing low-field MRI image quality.
- To develop a method for generating realistic noisy-clean datasets for training denoising models.
- To assess the effectiveness of NND in improving signal-to-noise ratio and preserving image details.
Main Methods:
- NND leverages inherent noise patterns from low-field images by analyzing corner patches.
- A paired noisy-clean dataset was created by adding synthesized noise to high-field images.
- A U-Net based denoising autoencoder was trained on the synthesized dataset and evaluated on diverse low-field MRI data.
Main Results:
- NND achieved significant signal-to-noise ratio (SNR) improvements: 32.76% (M4Raw), 19.02% (in vivo), and 8.16% (phantom).
- Qualitative analysis confirmed NND's superior preservation of structural details and edges compared to random noise denoising (RND).
- The method demonstrated robustness across different low-field MRI datasets and imaging conditions.
Conclusions:
- NND effectively mitigates noise in low-field MRI, enhancing diagnostic image quality.
- The approach improves the clinical utility of low-field MRI scanners, particularly in resource-constrained environments.
- This work facilitates broader access to advanced diagnostic imaging globally by overcoming key technical limitations.
More Related Videos
Related Concept Videos
NMR Spectrometers: Resolution and Error Correction
Atomic Nuclei: Magnetic Resonance
Atomic Nuclei: Nuclear Relaxation Processes
Magnetic Resonance Imaging
Diamagnetic Shielding of Nuclei: Local Diamagnetic Current
Double Resonance Techniques: Overview
Spin decoupling is usually achieved by...

