Systems and methods for compressed sensing for multi-shell magnetic resonance imaging

• US 9,208,587 B2
• Filed: 04/25/2014
• Issued: 12/08/2015
• Est. Priority Date: 04/25/2014
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Hardy, Christopher Judson, Marinelli, Luca, Menzel, Marion Irene, Tan, Ek Tsoon, Sperl, Jonathan Immanuel
• Primary Examiner(s): Lu, Tom Y
• Application Number:
  US14/261,623
• Publication Number:
  US 20150310640A1
• Time in Patent Office: 592 Days
• US Class Current: 1/1
• CPC Class Codes: G01R 33/4826 in three dimensions
• CPC Class Codes: G01R 33/5611 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 Class Codes: G01R 33/56341 Diffusion imaging
• CPC Class Codes: G06T 11/006 Inverse problem, transformation from projection-space into object-space, e.g. transform methods, back-projection, algebraic methods
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/20056 Discrete and fast Fourier transform, [DFT, FFT]
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A method of compressed sensing for multi-shell magnetic resonance imaging includes obtaining magnetic resonance imaging data, the data being sampled along multi-shell spherical coordinates, the spherical coordinates coincident with a plurality of spokes that converge at an origin, constructing a symmetric shell for each respective sampled multi-shell to create a combined set of data, performing a three-dimensional Fourier transform on the combined set of data to reconstruct an image, and de-noising the reconstructed image by iteratively applying a sparsifying transform on non-sampled data points of neighboring shells. The method can also include randomly under-sampling the imaging data to create missing data points. A system configured to implement the method and a non-transitory computer readable medium are also disclosed.

Claims

1. A method of compressed sensing for multi-shell magnetic resonance imaging, the method comprising:

providing a magnetic resonance imaging system including a control processor, gradient amplifiers, a radio frequency amplifier, and a pulse generator, wherein the control processor is in communication with the gradient amplifiers, the radio frequency amplifier, and the pulse generator, and wherein the control processor executes computer-readable instructions that cause the control processor to perform the steps of;

obtaining magnetic resonance imaging data, the data being sampled along multiple concentric spherical shells, the spherical shells coincident with a plurality of spokes that converge at an origin;

constructing a symmetric shell for each respective sampled shell to create a combined set of data;

performing a three-dimensional Fourier transform on the combined set of data to reconstruct an image; and

de-noising the reconstructed image by iteratively applying a sparsifying transform on non-sampled data points of neighboring shells.

2. The method of claim 1, including randomly under-sampling the imaging data to create missing data points.

3. The method of claim 2, including synthesizing the missing data points by applying prior knowledge about the under-sampled data points.

4. The method of claim 1, the constructing step including mirroring sampled data coincident with at least a portion of the plurality of sampled spokes along a spoke across the origin from the respective sampled spokes.

5. The method of claim 1, the de-noising step including applying a total variation operator by using one of an iterative shrinkage and thresholding algorithm and wavelet decomposition.

6. A non-transitory computer readable medium having stored thereon instructions which when executed by a processor cause the processor to perform a method of multi-shell magnetic resonance imaging, the method comprising:

obtaining magnetic resonance imaging data, the data being sampled along multiple concentric spherical shells, the spherical shells coincident with a plurality of spokes that converge at an origin;

constructing a symmetric shell for each respective sampled shell to create a combined set of data;

performing a three-dimensional Fourier transform on the combined set of data to reconstruct an image; and

de-noising the reconstructed image by iteratively applying a sparsifying transform on non-sampled data points of neighboring shells.

7. The medium of claim 6, including instructions to cause the processor to perform the step of randomly under-sampling the imaging data to create missing data points.

8. The medium of claim 7, including instructions to cause the processor to perform the step of synthesizing the missing data points by applying prior knowledge about the under-sampled data points.

9. The medium of claim 6, including instructions to cause the processor to perform the constructing step by including mirroring sampled data coincident with at least a portion of the plurality of sampled spokes along a spoke across the origin from the respective sampled spokes.

10. The medium of claim 6, including instructions to cause the processor to perform the de-noising step by including applying a total variation operator by using one of an iterative shrinkage and thresholding algorithm and wavelet decomposition.

11. A system for compressed sensing for multi-shell magnetic resonance imaging, the system comprising:

a magnetic resonance imaging system including a control processor, gradient amplifiers, a radio frequency amplifier, and a pulse generator, wherein the control processor is in communication with the gradient amplifiers, the radio frequency amplifier, and the pulse generator;

the control processor configured to execute computer-readable instructions that cause the control processor to perform a method including;

obtain magnetic resonance imaging data, the data being sampled along multiple concentric spherical shells, the spherical shells coincident with a plurality of spokes that converge at an origin;

construct a symmetric shell for each respective sampled shell to create a combined set of data;

perform a three-dimensional Fourier transform on the combined set of data to reconstruct an image; and

de-noise the reconstructed image by iteratively applying a sparsifying transform on non-sampled data points of neighboring shells.

12. The system of claim 11, including instructions to cause the control processor to randomly under-sample the imaging data to create missing data points.

13. The system of claim 12, including instructions to cause the control processor to synthesize the missing data points by applying prior knowledge about the under-sampled data points.

14. The system of claim 11, including instructions to cause the control processor to perform the constructing step by mirroring sampled data coincident with at least a portion of the plurality of sampled spokes along a spoke across the origin from the respective sampled spokes.

15. The system of claim 11, including instructions to cause the control processor to perform the de-noising step by the including applying a total variation operator by using one of an iterative shrinkage and thresholding algorithm and wavelet decomposition.