Stacked U-Nets with Self-Assisted Priors Towards Robust Correction of Rigid Motion Artifact in Brain MRI

Al‐masni, Mohammed A.; Lee, Seul; Yi, Jaeuk; Kim, Sewook; Gho, Sung-Min; Choi, Young Hun; Kim, Dong-Hyun

doi:10.48550/arxiv.2111.06401

Search citation statements

Order By: Relevance

Paper Sections

Select...

Introduction2

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2022

Publication Types

Select...

Other1

Relationship

Self Cite0

Independent1

Authors

Journals

Cited by 1 publication

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the advent of deep learning, the optimization methods were somewhat substituted by the convolutional neural networks (CNNs) [9,1,13], improving convergence of the non-convex motion parameter optimization procedure and/or increasing the reconstruction quality. A plethora of CNN-based deblurring methods, such as popular DnCNN [37], have then been implemented [29,30], without taking the physical nature of the MRI artifacts into account.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Autofocusing+: Noise-Resilient Motion Correction in Magnetic Resonance Imaging

Kuzmina¹,

Razumov²,

Rogov³

et al. 2022

Preprint

View full text Add to dashboard Cite

Image corruption by motion artifacts is an ingrained problem in Magnetic Resonance Imaging (MRI). In this work, we propose a neural network-based regularization term to enhance Autofocusing, a classic optimization-based method to remove motion artifacts. The method takes the best of both worlds: the optimization-based routine iteratively executes the blind demotion and deep learning-based prior penalizes for unrealistic restorations and speeds up the convergence. We validate the method on three models of motion trajectories, using synthetic and real noisy data. The method proves resilient to noise and anatomic structure variation, outperforming the state-of-the-art demotion methods.

show abstract

Section: Introductionmentioning

confidence: 99%

“…In the majority of works on MRI demotion, the corrupted data are either private [13,30] or generated by a model [1,29,5]. Herein, we also resort to the latter and employ a physics-based model to introduce the artifacts.…”

Section: Introductionmentioning

confidence: 99%