Multiple Slice k-space Deep Learning for Magnetic Resonance Imaging Reconstruction

Du, Tao; Zhang, Yanci; Shi, Xiaotong; Chen, Shuang

doi:10.1109/embc44109.2020.9175642

Cited by 17 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, several studies demonstrate that multi‐slice MR reconstruction offers the potential to exploit the similarities in image contents and phase across adjacent slices 27,48,49 . The multi‐slice acquisition allows adjacent slices to be acquired with different sampling patterns, which can provide complementary information between slices.…”

Section: Discussionmentioning

confidence: 99%

“…Recently, several studies demonstrate that multi-slice MR reconstruction offers the potential to exploit the similarities in image contents and phase across adjacent slices. 27,48,49 The multi-slice acquisition allows adjacent slices to be acquired with different sampling patterns, which can provide complementary information between slices. Motivated by these works, we performed a preliminary study to extend the DL-PF method by using multiple consecutive slices in PF reconstruction of typical multi-slice data (MSDL-PF).…”

Section: Joint Multi-slice Dl-pf Reconstruction Via Complementary Sam...mentioning

confidence: 99%

See 1 more Smart Citation

Partial Fourier reconstruction of complex MR images using complex‐valued convolutional neural networks

Xiao

Liu

et al. 2021

Magnetic Resonance in Med

View full text Add to dashboard Cite

Purpose To provide a complex‐valued deep learning approach for partial Fourier (PF) reconstruction of complex MR images. Methods Conventional PF reconstruction methods, such as projection onto convex sets (POCS), uses low‐resolution image phase information from the central symmetrically sampled k‐space for image reconstruction. However, this smooth phase constraint undermines the phase estimation accuracy in presence of rapid local phase variations, causing image artifacts and limiting the extent of PF reconstruction. Using both magnitude and phase characteristics in big complex image datasets, we propose a complex‐valued deep learning approach with an unrolled network architecture for PF reconstruction that iteratively reconstructs PF sampled data and enforces data consistency. We evaluate our approach for reconstructing both spin‐echo and gradient‐echo data. Results The proposed method outperformed the iterative POCS PF reconstruction method. It produced better artifact suppression and recovery of both image magnitude and phase details in presence of local phase changes. No noise amplification was observed even for highly PF reconstruction. Moreover, the network trained on axial brain data could reconstruct sagittal and coronal brain and knee data. This method could be extended to 2D PF reconstruction and joint multi‐slice PF reconstruction. Conclusion Our proposed method can effectively reconstruct MR data even at low PF fractions, yielding high‐fidelity magnitude and phase images. It presents a valuable alternative to conventional PF reconstruction, especially for phase‐sensitive 2D or 3D MRI applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Joint Multi-slice Dl-pf Reconstruction Via Complementary Sam...mentioning

confidence: 99%

Partial Fourier reconstruction of complex MR images using complex‐valued convolutional neural networks

Xiao

Liu

et al. 2021

Magnetic Resonance in Med

View full text Add to dashboard Cite

show abstract

“…In the field of MRI in the imaging industry, the data space is called k-space, that is, the Fourier frequency space. Each MRI image will have unique k-space data corresponding to it [14]. e Km and Kn directions of the k-space represent the frequency encoding direction and the phase encoding direction, respectively, and the gray value of each pixel on them determines the intensity of the magnetic resonance signal at that point.…”

Section: Principles Of Mrimentioning

confidence: 99%

Diagnostic Value of Magnetic Resonance Image Feature Analysis under Reconstruction Algorithm for Knee Epiphyseal Injury

Tang

Zhou

et al. 2022

Scientific Programming

View full text Add to dashboard Cite

To analyze the diagnostic value of magnetic resonance imaging (MRI) in epiphyseal injury of adolescent children, MRI images of 26 adolescent knee joint injuries were selected. The image display before and after reconstruction based on the compressed sensing theory (CS) algorithm was compared and analyzed, so did the signal-to-noise ratio of knee MRI reconstructed images under 2D random sampling and radial sampling at different sampling rates. The results showed that the sharpness, specificity, and accuracy of images reconstructed by the CS algorithm were significantly higher than those before reconstruction P < 0.05 . When the sampling rates were almost the same (20%, 40%, and 60%), the signal-to-noise ratio (SNR) of the reconstructed knee MRI images by random sampling was higher than that by radial sampling (about 9%, 6%, and 3% higher, respectively), that is, the information contained by random sampling was larger than that by radial sampling. It overcame the defects of undisplaced epiphyseal injuries in children and adolescents that cannot be found by plain X-ray films and can clearly and accurately diagnose epiphyseal injuries and fracture types that cannot be seen by plain X-ray films. To sum up, the MRI image reconstruction algorithm based on compressed sensing theory can effectively improve the diagnosis effect of knee epiphyseal injury.

show abstract

“…The summary of the NEUS algorithm is in the Algorithm 1. Some of the related works can be seen in,Zhu [2020],Huang et al [2019, 2020] Hu et al [2021,Lu et al [2017],Zhang et al [2021],Du et al [2020],Xu et al [2019Xu et al [ , 2020a,Zhao et al [2020],Kong et al [2018],Zhu et al [2020.…”

mentioning

confidence: 99%

A News-based Machine Learning Model for Adaptive Asset Pricing

Zhu¹,

Wu²,

Wells³

2021

Preprint

View full text Add to dashboard Cite

The paper proposes a new asset pricing model -the News Embedding UMAP Selection (NEUS) model, to explain and predict the stock returns based on the financial news. Using a combination of various machine learning algorithms, we first derive a company embedding vector for each basis asset from the financial news. Then we obtain a collection of the basis assets based on their company embedding.After that for each stock, we select the basis assets to explain and predict the stock return with highdimensional statistical methods. The new model is shown to have a significantly better fitting and prediction power than the Fama-French 5-factor model.

show abstract

Multiple Slice k-space Deep Learning for Magnetic Resonance Imaging Reconstruction

Cited by 17 publications

References 15 publications

Partial Fourier reconstruction of complex MR images using complex‐valued convolutional neural networks

Partial Fourier reconstruction of complex MR images using complex‐valued convolutional neural networks

Diagnostic Value of Magnetic Resonance Image Feature Analysis under Reconstruction Algorithm for Knee Epiphyseal Injury

A News-based Machine Learning Model for Adaptive Asset Pricing

Contact Info

Product

Resources

About