A Fast Low Rank Hankel Matrix Factorization Reconstruction Method for Non-Uniformly Sampled Magnetic Resonance Spectroscopy

Guo, Di; Lu, Hengfa; Qu, Xiaobo

doi:10.1109/access.2017.2731860

Cited by 36 publications

(32 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, the current implementation is computationally expensive, as it requires approximately 6 h to reconstruct each dataset on a computer cluster with processing speed of 2.3 GHz, 24 cores, and 128 GB of RAM using a supercomputer (http://www.sharcnet.ca). These obstacles may be overcome by implementing recent variations in the reconstruction algorithm …”

Section: Discussionmentioning

confidence: 99%

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Santos‐Díaz

Harasym

Noseworthy

2019

Magnetic Resonance in Med

View full text Add to dashboard Cite

Purpose Dynamic phosphorus MR spectroscopic imaging (31P‐MRSI) experiments require temporal resolution on the order of seconds to concurrently assess different muscle groups. A highly accelerated pulse sequence combining flyback echo‐planar spectroscopic imaging (EPSI) and compressed sensing was developed and tested in a phantom and healthy humans during an exercise‐recovery challenge of the lower leg muscles, using a clinical 3T MRI. Methods A flyback EPSI readout designed to achieve 2.25 × 2.25 cm2 resolution over a 18 × 18 cm2 field of view (i.e., 8 × 8 matrix) was combined with compressed sensing through the inclusion of pseudorandom gradient blips to sub‐sample the ky‐kt dimensions by a factor of 2.7×, achieving a temporal resolution of 9 s. The sequence was first tested in a phantom to assess performance compared to fully sampled EPSI (fidEPSI) and phase encoded chemical shift imaging (fidCSI). Then, tests were performed in 11 healthy volunteers during an exercise‐recovery challenge of the lower leg muscles. Voxels containing tissue from different muscle groups were evaluated measuring percentage phosphocreatine (%PCr) depletion, time constant of PCr recovery (τPCr) and intracellular pH at rest and following exercise. Results The sequence was capable to track the dynamic PCr response of multiple muscles simultaneously. No statistical differences were found in the metabolite ratio, pH or linewidth when compared with fidEPSI and fidCSI in the phantom study. Dynamic experiments showed differences in PCr depletion when comparing soleus with gastrocnemius muscles. Intracellular pH, τPCr and %PCr decrease were consistent with reported values. Conclusion Highly accelerated 31P‐MRSI combining flyback EPSI and compressed sensing is capable of assessing concurrent energy metabolism in multiple muscle groups using a clinical 3T MR system.

show abstract

Section: Discussionmentioning

confidence: 99%

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Santos‐Díaz

Harasym

Noseworthy

2019

Magnetic Resonance in Med

View full text Add to dashboard Cite

show abstract

“…where H is an operator that converts M into a block Hankel matrix [35]- [38] with a dimension of (k 1 k 2 ) × ((N − k 1 + 1)(M − k 2 + 1)), and k 1 and k 2 are pencil parameters.…”

Section: B Alohamentioning

confidence: 99%

“…These observations imply that both the low-rankness and estimated self-consistency improve the reconstruction in the proposed STDLR-SPIRiT. To tackle (18), time-consuming singular value decomposition (SVD) is hardly avoided since singular value thresholding is utilized for nuclear norm minimization [37], [38], [40]. In our case, the size of low-rank matrix cascaded of Hankel matrices (Jk 1 k 2 ) × ((N − k 1 + 1)(M − k 2 + 1)) increases dramatically after pencil parameters k 1 , k 2 increase, resulting in a sharp increase in SVD calculation time.…”

Section: A Proposed Reconstruction Modelmentioning

confidence: 99%

“…Thus, we seek an SVD-free algorithm to reduce the computation time. Here, we employ matrix factorization technique [37], [38], [40], leading to a dramatic reduction of computation time. The SVD-free algorithm is based on the relationship [39] shown below:…”

Section: A Proposed Reconstruction Modelmentioning

confidence: 99%

“…Here, we adopt the Alternating Direction Method of Multiplier (ADMM) [41] to deal with the proposed model in (20). It should be noticed that the non-convex optimization problem (20) can be successfully solved by ADMM [31], [37], [38]. The augmented Lagrangian form of (20) is max Di min X,Pi,Qi…”

Section: B Svd-free Low Rank Reconstruction Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

Image reconstruction with low-rankness and self-consistency of k-space data in parallel MRI

Zhang

Guo

Huang

et al. 2020

Medical Image Analysis

Self Cite

View full text Add to dashboard Cite

Parallel magnetic resonance imaging has served as an effective and widely adopted technique for accelerating scans. The advent of sparse sampling offers aggressive acceleration, allowing flexible sampling and better reconstruction. Nevertheless, faithfully reconstructing the image from limited data still poses a challenging task. Recent low-rank reconstruction methods exhibit superiority in providing a high-quality image. However, none of them employ the routinely acquired calibration data for improving image quality in parallel magnetic resonance imaging. In this work, an image reconstruction approach named STDLR-SPIRiT was proposed to explore the simultaneous twodirectional low-rankness (STDLR) in the k-space data and to mine the data correlation from multiple receiver coils with the iterative self-consistent parallel imaging reconstruction (SPIRiT). The reconstruction problem was then solved with a singular value decomposition-free numerical algorithm. Experimental results of phantom and brain imaging data show that the proposed method outperforms the state-of-the-art methods in terms of suppressing artifacts and achieving the lowest error. Moreover, the proposed method exhibits robust reconstruction even when the autocalibration signals are limited in parallel imaging. Overall the proposed method can be exploited to achieve better image quality for accelerated parallel magnetic resonance imaging.

show abstract

Accelerated Nuclear Magnetic Resonance Spectroscopy with Deep Learning

Huang

et al. 2020

Angewandte Chemie

Self Cite

View full text Add to dashboard Cite

Nuclear magnetic resonance (NMR) spectroscopy serves as an indispensable tool in chemistry and biology but often suffers from long experimental times. We present a proof‐of‐concept of the application of deep learning and neural networks for high‐quality, reliable, and very fast NMR spectra reconstruction from limited experimental data. We show that the neural network training can be achieved using solely synthetic NMR signals, which lifts the prohibiting demand for a large volume of realistic training data usually required for a deep learning approach.

show abstract

A Fast Low Rank Hankel Matrix Factorization Reconstruction Method for Non-Uniformly Sampled Magnetic Resonance Spectroscopy

Cited by 36 publications

References 27 publications

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Image reconstruction with low-rankness and self-consistency of k-space data in parallel MRI

Accelerated Nuclear Magnetic Resonance Spectroscopy with Deep Learning

Contact Info

Product

Resources

About

A Fast Low Rank Hankel Matrix Factorization Reconstruction Method for Non-Uniformly Sampled Magnetic Resonance Spectroscopy

Cited by 36 publications

References 27 publications

Dynamic 31P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Dynamic 31P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Image reconstruction with low-rankness and self-consistency of k-space data in parallel MRI

Accelerated Nuclear Magnetic Resonance Spectroscopy with Deep Learning

Contact Info

Product

Resources

About

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing

Dynamic ³¹P spectroscopic imaging of skeletal muscles combining flyback echo‐planar spectroscopic imaging and compressed sensing