Related Experiment Video
Updated: May 28, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.6K
3D Wasserstein Generative Adversarial Network with Dense U-Net-Based Discriminator for Preclinical fMRI Denoising
Sima Soltanpour1, Arnold Chang2, Dan Madularu3,4
1School of Information Technology, Carleton University, 1125 Colonel By Dr, Ottawa, Ontario, K1S 5B6, Canada. simasoltanpour@cunet.carleton.ca.
Journal of Imaging Informatics in Medicine
|February 12, 2025
Summary
This study introduces 3D U-WGAN, a novel algorithm for denoising preclinical functional magnetic resonance imaging (fMRI) data. The method enhances image quality and signal-to-noise ratio while preserving crucial brain structures.
Area of Science:
- Neuroimaging
- Biomedical Engineering
- Artificial Intelligence
Background:
- Functional magnetic resonance imaging (fMRI) is vital for studying brain function but is susceptible to noise.
- Denoising fMRI data is critical, especially for preclinical studies facing unique challenges like low signal-to-noise ratios and anatomical variability.
Purpose of the Study:
- To develop a structure-preserved algorithm for effective denoising of preclinical fMRI data.
- To improve the signal-to-noise ratio and image quality of fMRI scans without compromising structural integrity.
Main Methods:
- Proposed a novel 3D Wasserstein generative adversarial network (3D U-WGAN) with a 3D dense U-Net discriminator.
- Utilized a 4D data configuration for comprehensive spatio-temporal denoising.
- Incorporated adversarial loss and feature space distance measurements to prevent oversmoothing and enhance perceptual similarity.
Main Results:
- 3D U-WGAN significantly improved image quality in both resting-state and task-based preclinical fMRI data.
- The algorithm effectively enhanced signal-to-noise ratios while minimizing structural alterations.
- Demonstrated superior performance compared to state-of-the-art methods on simulated and real preclinical fMRI datasets.
Conclusions:
- The proposed 3D U-WGAN is a powerful tool for denoising preclinical fMRI data, offering improved image quality and structural preservation.
- This method advances the analysis of preclinical fMRI, enabling more reliable insights into brain function.
