Abstract:Magnetic Resonance Imaging (MRI) is sensitive to motion caused by patient movement due to the relatively long data acquisition time. This could cause severe degradation of image quality and therefore affect the overall diagnosis. In this paper, we develop an efficient retrospective 2D deep learning method called stacked U-Nets with self-assisted priors to address the problem of rigid motion artifacts in 3D brain MRI. The proposed work exploits the usage of additional knowledge priors from the corrupted images themselves without the need for additional contrast data. The proposed network learns the missed structural details through sharing auxiliary information from the contiguous slices of the same distorted subject. We further design a refinement stacked U-Nets that facilitates preserving the spatial image details and improves the pixel-to-pixel dependency. To perform network training, simulation of MRI motion artifacts is inevitable. The proposed network is optimized by minimizing the loss of structural similarity (SSIM) using the synthesized motion-corrupted images from 83 real motion-free subjects. We present an intensive analysis using various types of image priors: the proposed self-assisted priors and priors from other image contrast of the same subject. The experimental analysis proves the effectiveness and feasibility of our self-assisted priors since it does not require any further data scans. The overall image quality of the motion-corrected images via the proposed motion correction network significantly improves SSIM from 71.66% to 95.03% and declines the mean square error from 99.25 to 29.76. These results indicate the high similarity of the brain's anatomical structure in the corrected images compared to the motion-free data. The motion-corrected results of both the simulated and real motion data showed the potential of the proposed motion correction network to be feasible and applicable in clinical practices.

Multi-Scale Network with the Deeper and Wider Residual Block for MRI Motion Artifact Correction

CMR Motion Artifact Correction Using Generative Adversarial Nets

Cine Cardiac MRI Motion Artifact Reduction Using a Recurrent Neural Network

Motion artifact reduction for magnetic resonance imaging with deep learning and k-space analysis

Iterative Learning for Joint Image Denoising and Motion Artifact Correction of 3D Brain MRI

Motion Correction of High-Resolution Spatiotemporally Encoded MRI Based on Deep Learning

Reconstruction of Cardiac Cine MRI Using Motion-Guided Deformable Alignment and Multi-Resolution Fusion

Learning residual motion correction for fast and robust 3D multiparametric MRI

Reconstruction of Cardiac Cine MRI Using Motion‐Guided Deformable Alignment and Multi‐Resolution Fusion

A foundation model for enhancing magnetic resonance images and downstream segmentation, registration and diagnostic tasks

Respiratory Motion Correction in Abdominal MRI using a Densely Connected U-Net with GAN-guided Training

Deep Residual Learning for Accelerated MRI using Magnitude and Phase Networks

Motion-Informed Deep Learning for Brain MR Image Reconstruction Framework

A cardiac MRI motion artifact reduction method based on edge enhancement network

Dual Domain Motion Artifacts Correction for MR Imaging Under Guidance of K-space Uncertainty

Deep Network Cascade for Dynamic Cardiac MRI Reconstruction With Motion Feature Incorporation and the Fourier Neural Attention

Stop moving: MR motion correction as an opportunity for artificial intelligence

Deep unregistered multi-contrast MRI reconstruction

Performance of a GPU- and Time-Efficient Pseudo 3D Network for Magnetic Resonance Image Super-Resolution and Motion Artifact Reduction

Stacked U-Nets with self-assisted priors towards robust correction of rigid motion artifact in brain MRI

DCT-net: Dual-domain cross-fusion transformer network for MRI reconstruction