System for reconstructing MRI images acquired in parallel

1. A system for parallel image processing in MR imaging, comprising:

a plurality of MR imaging RF coils for individually receiving MR imaging data representing a slice of patient anatomy;

an MR imaging system for using the plurality of RF coils for acquiring a corresponding plurality of image data sets of said slice; and

an image data processor comprising at least one processing device configured to:

derive a first set of weights for generating a calibration data set comprising a subset of k-space data of composite image data representing said plurality of image data sets,

generate a first MR image data set representative of the calibration data set, wherein the first MR image data set is generated by reconstructing an initial image by convolving the first set of weights with the k-space data and combining the convolved results across the plurality of image data sets;

derive a second set of weights using said calibration data set and a sparsifying transform of the generated first MR image data set, and

use said second set of weights to generate a second MR image data set representing a single image having a reduced set of data components relative to said first composite MR image data set.

2. A system according to claim 1 , wherein

said image data processor generates said first MR image data set representing a single image using said first set of weights by performing a weighted combination of luminance representative data of individual corresponding pixels of said plurality of image data sets in providing an individual pixel luminance value of said first MR image data set.

3. A system according to claim 1 , wherein

said image data processor derives said second set of weights using said calibration data set and the generated first MR image data set by reducing noise in said generated first MR image data set by generating a reduced set of significant components in a predetermined transform domain representation of data representing the first MR image data set to provide a de-noised composite MR image data set.

4. A system according to claim 1 , wherein

said image data processor derives said second set of weights using said calibration data set and the generated first MR image data set by using an inverse Fourier transform and a wavelet transform in iteratively converting between the weights and a wavelet representation of interpolated image data in providing a compressed sensing sparse transformation of the weights.

5. A system according to claim 1 , wherein

said image data processor generates said first MR image data set by deriving said first set of weights using said plurality of image data sets of said slice.

6. A system according to claim 1 , wherein

said plurality of image data sets of said slice individually comprise a reduced size block of calibration k-space data.

7. A system according to claim 1 , wherein

said image data processor uses weights to interpolate missing components in an acquired undersampled k-space data set from a linear combination of k-space data points for image data from an individual RF coil.

8. A system according to claim 1 , wherein

said image data processor generates said first set of weights using a fitting function to fit the acquired subset of k-space data.

9. A system according to claim 1 , including

an image generator comprising at least one processing device conditioned for generating an MR image derived using said second MR image data set.

10. A system according to claim 1 , wherein

said second MR image data set has a reduced set of data components comprising a data set of increased sparsity relative to said first MR image data set.

11. A system according to claim 10 , wherein

said reduced set of components comprise components in a predetermined transform domain representation of data representing a composite image.

12. A system according to claim 1 , wherein

said image data processor uses the weights to provide a weighted linear combination of luminance representative data of individual corresponding pixels of said plurality of image data sets.

13. A system according to claim 1 , wherein

said image data processor in generating an MR image data set performs an image reconstruction.

14. A system according to claim 1 , wherein

said MR imaging system uses the plurality of RF coils for acquiring a corresponding plurality of image data sets of said slice by using a parallel imaging method compatible with at least one of (a) SMASH (SiMultaneous Acquisition of SpatialHarmonics) and (b) GRAPPA (Gene-Ralized Autocalibrating Partially Parallel Acquisition).

15. A system for parallel image processing in MR imaging, comprising:

a plurality of MR imaging RF coils for individually receiving MR imaging data representing a slice of patient anatomy;

an MR imaging system configured to use the plurality of RF coils to acquire a corresponding plurality of image data sets of said slice; and

an image data processor comprising at least one processing device configured to:

derive a set of weights for generating a calibration data set comprising a subset of k-space data of composite image data representing said plurality of image data sets,

iteratively update the set of weights by a process until a convergence condition is satisfied, the process comprising:

using said calibration data set and the set of weights to generate an initial MR image data set, wherein the initial MR image data set is generated by reconstructing an initial image by convolving the set of weights with the k-space data and combining the convolved results across the plurality of image data sets, and

updating the set of weights using a generated calibration data set and a sparsifying transform of the generated initial MR image data set; and

after the convergence condition is satisfied, using said further updated set of weights in generating a further MR image data set representing a single image having a reduced set of data components relative to a previously generated initial MR image data set.

16. A method for parallel image processing in MR imaging, comprising the activities of:

using a plurality of RF coils for acquiring a corresponding plurality of image data sets of an image slice;

deriving a first set of weights for generating a calibration data set comprising a subset of k-space data of composite image data representing said plurality of image data sets;

using said calibration data set to generate a first MR image data set wherein the first MR image data set is generated by reconstructing an initial image by convolving the set of weights with the k-space data and combining the convolved results across the plurality of image data sets;

updating the set of weights using said calibration data set and a sparsifying transform of the generated first MR image data set; and

using said updated set of weights to generate a second MR image data set representing a single image having a reduced set of data components relative to said first MR image data set.

17. A method according to claim 16 , including the activity of:

generating said first MR image data set representing a single image using said first set of weights by performing a weighted combination of luminance representative data of individual corresponding pixels of said plurality of image data sets in providing an individual pixel luminance value of said first MR image data set.

18. A method for parallel image processing in MR imaging, comprising the activities of:

using a plurality of RF coils for acquiring a corresponding plurality of image data sets of an image slice;

deriving a set of weights for generating a calibration data set comprising a subset of k-space data of composite image data representing said plurality of image data sets;

iteratively updating the set of weights by a process until a convergence condition is satisfied, the process comprising:

using said calibration data set to generate an initial MR image data set, wherein the initial MR image data set is generated by reconstructing an initial image by convolving the set of weights with the k-space data and combining the convolved results across the plurality of image data sets, and

updating the set of weights using a generated calibration data set and a sparsifying transform of the generated initial MR image data set; and

after the convergence condition is satisfied, using said updated set of weights in generating a further MR image data set representing a single image having a reduced set of data components relative to a previously generated initial MR image data set.