Self consistent parallel imaging with temporal sensitivity estimation in real-time magnetic resonance imaging

Patent metadata

Field	Value
Publication number	US-9279873-B2
Application number	US-201213534112-A
Country	US
Kind code	B2
Filing date	Jun 27, 2012
Priority date	Jul 28, 2011
Publication date	Mar 8, 2016
Grant date	Mar 8, 2016

Abstract

Parallel imaging magnetic resonance reconstruction is performed with temporal sensitivity. Rather than estimate the coil sensitivity once for each coil of an array, the coil sensitivity at different times is estimated. The movement of the patient may result in different sensitivities at different times. By using the time varying sensitivity in iterative, self-consistent, non-linear parallel imaging, real-time imaging may be provided with stable artifacts in view of increasing SNR even with higher reduction factors (e.g., 4-6).

First claim

We claim:

1. A method for parallel imaging with temporal sensitivity in magnetic resonance reconstruction, the method comprising:
acquiring, with multiple coils, magnetic resonance data for parallel imaging over multiple phases of a physiological cycle, the magnetic resonance data representing a dynamic interior region of a patient;
performing, for each of the multiple coils and each phase, an initial reconstruction;
temporally filtering the initial reconstructions for each of the multiple coils;
estimating a coil sensitivity for each of the multiple coils at each of the multiple phases from results of the temporally filtering;
solving a non-linear reconstruction for each of the phases as a function of the coil sensitivities and the magnetic resonance data; and
generating, from an output of the solving, a sequence of images representing the dynamic interior region.

2. The method of claim 1 wherein acquiring comprises acquiring the magnetic resonance data as interleaved with a reduction factor of four, five, or six.

3. The method of claim 1 wherein performing the initial reconstruction comprises performing a linear reconstruction of k-space data.

4. The method of claim 1 wherein temporally filtering comprises applying a Karhunen-Loeve transform.

5. The method of claim 4 wherein estimating from the results of the temporally filtering comprise estimating from only a first three modes output by the Karhunen-Loeve transform.

6. The method of claim 4 wherein performing the initial reconstruction comprises performing generalized auto calibrating partially parallel acquisitions reconstruction.

7. The method of claim 1 wherein estimating comprises estimating with an iterative self-constraint parallel imaging reconstruction calibration.

8. The method of claim 1 wherein solving the non-linear reconstruction comprises:
applying a least square matrix inversion solver; and
applying a non-linear conjugate gradient solver.

9. The method of claim 1 wherein solving comprises generating object space data for each of the multiple phases using the coil sensitivities for the respective phase.

10. The method of claim 1 wherein generating comprises generating each of the images from the magnetic resonance data from the multiple coils.

11. In a non-transitory computer readable storage medium having stored therein data representing instructions executable by a programmed processor for parallel imaging with temporal sensitivity in magnetic resonance reconstruction, the storage medium comprising instructions for:
estimating coil sensitivities of an array of coils at different times from an outcome of linear reconstruction and filtering; and
performing iterative self-consistent parallel imaging reconstruction from k-space data received with the array of coils, the performing being a function of the coil sensitivities at the different times.

12. The non-transitory computer readable storage medium of claim 11 wherein estimating the coil sensitivities comprises estimating the coil sensitivities from regularly interleaved Cartesian k-space sampling.

13. The non-transitory computer readable storage medium of claim 11 further comprising:
linearly reconstructing data from interleaved k-space data for each coil of the array and for each of the different times;
temporally filtering the linearly reconstructed data across the different times for each of the coils; and
outputting the outcome from the temporally filtering.

14. The non-transitory computer readable storage medium of claim 13 wherein linearly reconstructing comprises performing generalized auto calibrating partially parallel acquisitions reconstruction and wherein temporally filtering comprises applying a Karhunen-Loeve transform, and wherein outputting comprises outputting only a first three modes of the Karhunen-Loeve transform.

15. The non-transitory computer readable storage medium of claim 13 wherein estimating the coil sensitivities comprises estimating kernels for the different times in calibration for the iterative self-consistent parallel imaging reconstruction.

16. The non-transitory computer readable storage medium of claim 13 wherein performing comprises performing with the coil sensitivities for each of the coils and each of the different times.

17. In a non-transitory computer readable storage medium having stored therein data representing instructions executable by a programmed processor for parallel imaging with temporal sensitivity in magnetic resonance reconstruction, the storage medium comprising instructions for:
receiving, for coils of an array, interleaved k-space data in frames representing different phases;
generating full k-space data for the frames;
filtering the full k-space data;
estimating self-consistent parallel imaging reconstruction kernels for the different phases from the filtered full k-space data; and
solving a non-linear reconstruction from the interleaved k-space data and the kernels.

18. The non-transitory computer readable storage medium of claim 17 wherein solving comprises solving with a linear matrix inversion with LSQR and then a non-linear conjugate gradient iteration.

19. The non-transitory computer readable storage medium of claim 17 wherein generating the full k-space data comprises linearly reconstructing the full k-space data into a Cartesian sampling, and wherein filtering comprises temporally filtering with a Karhunen-Loeve transform.

20. The non-transitory computer readable storage medium of claim 17 wherein estimating comprises estimating for each of the frames.

Assignees

• Xue Hui
• Ding Yu
• Siemens Corp

Inventors

• Xue Hui
• Ding Yu

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

• G01R33/56509

  due to motion, displacement or flow, e.g. gradient moment nulling (G01R33/567 takes precedence) · CPC title