Magnetic resonance imaging apparatus and control method for reconstruction of undersampled data

Patent metadata

Field	Value
Publication number	US-10317496-B2
Application number	US-201414501195-A
Country	US
Kind code	B2
Filing date	Sep 30, 2014
Priority date	Dec 10, 2013
Publication date	Jun 11, 2019
Grant date	Jun 11, 2019

Abstract

An MRI apparatus includes: a receive coil assembly including channels, and configured to receive an MR signal from an object; a data generator configured to generate undersampled image data on a k-space based on the MR signal; and a reconstructed image generator configured to generate a first reconstructed image from the undersampled image data using a parallel imaging method, and a second reconstructed image from the undersampled image data using compressed sensing. According to the MRI apparatus, since image reconstruction is performed using random undersampled image data, a parallel imaging method, and compressed sensing, it is possible to increase a speed of image acquisition and, at the same time, to improve the quality of images.

First claim

What is claimed is:

1. A magnetic resonance imaging (MRI) apparatus comprising:
a receive coil assembly including channels, and configured to receive an MR signal from an object;
a data generator configured to generate undersampled image data on a k-space based on the MR signal;
a reconstruction processor configured to generate a plurality of first reconstructed images from the undersampled image data using a parallel imaging technique, and generate a plurality of second reconstructed images using the plurality of first reconstructed images as initial values in dictionary learning compressed sensing, and combine the plurality of second reconstructed images into an image using a square sum or a complex sum, thereby generating a final reconstructed image;
a data storage configured to store the MR signal, the undersampled image data, the plurality of first reconstructed images, the plurality of second reconstructed images and the final reconstructed image; and
a display configured to display the final reconstructed image,
wherein the undersampled image data includes non-uniform random undersampled image data,
wherein the reconstruction processor is further configured to generate the plurality of second reconstructed images by using the dictionary learning compressed sensing by creating a global dictionary or an adaptive dictionary using a k-singular value decomposition (k-SVD), acquiring a sparse coefficient using an orthogonal matching pursuit (OMP), and performing patched reconstruction repeatedly, the dictionary learning compressed sensing being expressed by Equation (1):

min

x
,
D
,
Z

⁢

∑

i
,
j

⁢

⁢



R
ij

⁢
x

-

Dz
ij



2
2

+

λ
⁢



F
u

⁢
x

-
y



2
2

:



z
ij



0

≤

T
0

⁢

⁢

∀
i

,
j
,

where x represents each of the plurality of second reconstructed images,
y represents an original k-space corresponding to each of the plurality of second reconstructed images,
R ij represents a function for selecting a patch,
F u represents fast Fourier transform (FFT),
D represents the global dictionary or the adaptive dictionary acquired through the k-singular value decomposition (k-SVD),
Z ij represents the sparse coefficient acquired through the orthogonal matching pursuit (OMP),
II II 2 represents L2-norm,
II II 0 represents L0-norm,
λ represents a weight of the L2-norm, and
T 0 represents a degree of sparsity,
wherein the reconstruction processor is further configured to provide, as the plurality of second reconstructed images, images which satisfy a convergent condition of the Equation (1).

2. The MRI apparatus according to claim 1 , wherein the parallel imaging technique includes at least one from among of Sensitivity Profiles From an Array of Coils for Encoding and Reconstruction in Parallel (SPACE RIP), Simultaneous acquisition of spatial harmonics (SMASH), Partially Parallel Imaging With Localized Sensitivities (PILS), and Generalized Autocalibrating Partially Parallel Acquisitions (GRAPPA).

3. The MRI apparatus according to claim 1 , wherein the dictionary learning compressed sensing uses the global dictionary or the adaptive dictionary.

4. The MRI apparatus according to claim 1 , wherein the data generator is further configured to generate the undersampled image data by using at least one from among a Cartesian trajectory, a radial trajectory, and a spiral trajectory.

5. The MRI apparatus according to claim 1 , wherein the data generator is further configured to generate a plurality of image data in correspondence to the channels.

6. The MRI apparatus according to claim 5 , wherein the reconstruction processor is further configured to generate the plurality of first reconstructed images and the plurality of second reconstructed images in correspondence to the plurality of image data.

7. A control method of a magnetic resonance imaging (MRI) apparatus, the control method comprising:
receiving an MR signal from an object by a receive coil assembly including channels;
generating an undersampled image data on a k-space based on the MR signal by a data generator;
generating a plurality of first reconstructed images from the undersampled image data using a parallel imaging technique, by a reconstruction processor;
subsequently, generating a plurality of second reconstructed images using the plurality of first reconstructed images as initial values in dictionary learning compressed sensing, by the reconstruction processor;
combining the plurality of second reconstructed images into an image using a square sum or a complex sum, thereby generating a final reconstructed image, by the reconstruction processor;
storing the MR signal, the undersampled image data, the plurality of first reconstructed images, the plurality of second reconstructed images and the final reconstructed image in a data storage; and
displaying the final reconstructed image on a display,
wherein the

Assignees

• Samsung Electronics Co Ltd

Inventors

• Choi Sang Cheon
• Kim Joon Soo
• Seong Yeol Min
• Choi Yang Lim
• Gulaka Praveen

Classifications

• G01R33/5611 Primary

  Parallel magnetic resonance imaging, e.g. sensitivity encoding [SENSE], simultaneous acquisition of spatial harmonics [SMASH], unaliasing by Fourier encoding of the overlaps using the temporal dimension [UNFOLD], k-t-broad-use linear acquisition speed-up technique [k-t-BLAST], k-t-SENSE (structural details of arrays of sub-coils G01R33/3415) · CPC title