Related Experiment Video
Updated: Nov 4, 2025
06:52
Author Spotlight: Advancing 3D Cytoarchitecture Analysis - Rapid Volumetric Reconstruction of the Human Brain
Published on: January 26, 2024
2.5K
Denoising of 3D Brain MR Images with Parallel Residual Learning of Convolutional Neural Network Using Global and
Liang Wu1, Shunbo Hu2, Changchun Liu1
1School of Control Science and Engineering, Shandong University, Jinan 250061, China.
Computational Intelligence and Neuroscience
|May 31, 2021
Summary
This study introduces 3D-Parallel-RicianNet, a deep learning method for denoising magnetic resonance (MR) images. The novel approach effectively suppresses noise while preserving crucial image structures for improved medical diagnoses.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Image Processing
Background:
- Magnetic resonance (MR) images are susceptible to noise, hindering accurate medical diagnoses.
- Existing noise removal algorithms show promise but can be improved for complex MR image noise.
Purpose of the Study:
- To develop an advanced deep learning-based denoising method for MR images.
- To enhance the accuracy and reliability of MR image analysis for clinical applications.
Main Methods:
- Proposed 3D-Parallel-RicianNet, integrating dilated convolution residual (DCR) and depthwise separable convolution residual (DSCR) modules.
- Employed a parallel network architecture to fuse global and local features efficiently.
- Utilized a reconstruction module to generate clean MR images from noisy inputs.
Main Results:
- 3D-Parallel-RicianNet demonstrated superior performance over state-of-the-art methods on simulated and real MR datasets.
- Achieved higher peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).
- Showcased excellent noise suppression capabilities while preserving essential image structures.
Conclusions:
- The proposed 3D-Parallel-RicianNet effectively removes Rician noise from MR images.
- The method offers a significant advancement in MR image denoising for improved diagnostic accuracy.
- This deep learning approach provides a robust solution for preserving image quality in medical imaging.
