Method for exploiting structure in sparse domain for magnetic resonance image reconstruction

1. A method for constructing an image, comprising:

acquiring image data in a first domain;

transforming the acquired image data from the first domain into a second domain in which the acquired image data exhibits a higher degree of sparsity than exhibited in the first domain;

approximating an initial set of transform coefficients for transforming the acquired image data from the second domain into a third domain in which the image may be displayed;

updating the approximated initial set of transform coefficients based on context information pertaining to the approximate initial set of transform coefficients;

constructing an image in the third domain based on the updated set of transform coefficients; and

displaying the constructed image,

wherein the context information includes parent coefficients of the initial set of transform coefficients that has already been found within the image data in the second domain and a weighted sum of neighbor coefficients proximate to the initial set of transform coefficients that has already been found within the image data in the second domain, and

wherein for each initial transform coefficient that has already been found, the parent coefficient is weighed equally, within the context information, with the sum of neighbor coefficients.

2. The method of claim 1 , wherein the first domain is a sensing domain, the second domain is a sparsity domain, and the third domain is an image domain.

3. The method of claim 2 , wherein the sparsity domain is a wavelet transform domain.

4. The method of claim 2 , wherein the sparsity domain is a total-variation domain.

5. The method of claim 1 , wherein approximating the initial set of transform coefficients is performed using compressed sensing.

6. The method of claim 1 , wherein constructing the image in the third domain includes performing an inverse wavelet transform using the updated set of transform coefficients.

7. The method of claim 1 , wherein proximity, for determining the locations proximate to where initial transform coefficients have already been found, is measured along a diagonal orientation along subbands resulting from a decomposition of wavelet coefficients.

8. The method of claim 1 , wherein a weighing of where additional transform coefficients are likely to be located is based on the context information.

9. The method of claim 1 , wherein the acquired image data is acquired using a magnetic resonance imager.

10. The method of claim 1 , wherein the acquired image data is acquired using a computed tomography imager.

11. The method of claim 1 , wherein approximating an initial set of transform coefficients includes performing compressed sensing (CS) reconstruction.

12. The method of claim 11 , wherein performing compressed sensing (CS) reconstruction is performed in conjunction with Iterative Hard Thresholding (IHT), Orthogonal Matching Pursuit (OMP), Iteratively Reweighted Least Squares (IRLS), or reweighted l 1 minimization (RL1).

13. A method for constructing an image, comprising:

acquiring image data in a sensing domain, wherein the acquired image data is insufficient to construct an image using a Nyquist technique;

transforming the acquired image data from the sensing domain into a wavelet transform domain in which the acquired image data exhibits a higher degree of sparsity than exhibited in the sensing domain; and

constructing an image in an image domain based on the acquired image data transformed into the wavelet transform domain using compressed sensing (CS) reconstruction, wherein image construction is influenced by a maximization of sparsity of the acquired image data in the wavelet transform domain and a priori knowledge about locations of additional transform coefficients for the acquired image data within the wavelet transform domain; and

displaying the constructed image,

wherein said a priori knowledge about locations of additional transform coefficients for the acquired image data within the wavelet transform domain includes weighing context information pertaining to an approximate initial set of transform coefficients,

wherein the context information includes parent coefficients of the initial set of transform coefficients that has already been found within the image data in the wavelet transform domain and a weighted sum of neighbor coefficients proximate to the initial set of transform coefficients that has already been found within the image data in the wavelet transform domain, and

wherein for each initial transform coefficient that has already been found, the parent coefficient is weighed equally, within the context information, with the sum of neighbor coefficients.

14. The method of claim 13 , wherein proximity, for determining the locations proximate to where initial transform coefficients have already been found, is measured along diagonal orientation of subbands resulting from a decomposition of wavelet coefficients.

15. The method of claim 13 , wherein the acquired image data is acquired using a magnetic resonance imager or a computed tomography imager.

16. The method of claim 13 , wherein performing compressed sensing (CS) reconstruction is performed in conjunction with Iterative Hard Thresholding (IHT), Orthogonal Matching Pursuit (OMP), Iteratively Reweighted Least Squares (IRLS), or reweighted l 1 minimization (RL1).

17. A computer system comprising:

a processor; and

a non-transitory, tangible, program storage medium, readable by the computer system, embodying a program of instructions executable by the processor to perform method steps for constructing an image, the method comprising:

acquiring image data in a sensing domain;

transforming the acquired image data from the sensing domain into a wavelet transform domain in which the acquired image data exhibits a higher degree of sparsity than exhibited in the sensing domain;

approximating an initial set of transform coefficients for transforming the acquired image data from the wavelet transform domain into an image domain in which the image may be displayed by performing compressed sensing (CS) reconstruction;

updating the approximated initial set of transform coefficients based on context information pertaining to the approximate initial set of transform coefficients;

constructing an image in the image domain based on the updated set of transform coefficients; and

displaying the constructed image,

wherein the context information includes parent coefficients of the initial set of transform coefficients that has already been found within the image data in the wavelet transform domain and a weighted sum of neighbor coefficients proximate to the initial set of transform coefficients that has already been found within the image data in the wavelet transform domain, and

wherein for each initial transform coefficient that has already been found, the parent coefficient is weighed equally, within the context information, with the sum of neighbor coefficients.

18. The computer system of claim 17 , wherein approximating the initial set of transform coefficients is performed using compressed sensing.

19. The computer system of claim 17 , wherein the acquired image data is acquired using a magnetic resonance imager.

20. The computer system of claim 17 , wherein performing compressed sensing (CS) reconstruction is performed in conjunction with Iterative Hard Thresholding (IHT), Orthogonal Matching Pursuit (OMP), Iteratively Reweighted Least Squares (IRLS), or reweighted l 1 minimization (RL1).