Computer implemented method and system for magnetic resonance imaging

1. A computer implemented method for magnetic resonance imaging, the method comprising:

receiving a first subset of k-space data and a second subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

sampling the first subset of k-space data and the second subset of k-space data;

choosing the sampled first subset of k-space data as a base subset of k-space data;

estimating motion parameters of the second subset of k-space data against the base subset of k-space data;

acquiring a non-linear motion estimating function by function approximation by training a neural network; and

correcting the second subset of k-space data based on the motion parameters of the second subset of k-space data, wherein the motion parameters of the second subset of k-space data are parameters of the non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and receiving the second subset of k-space data.

2. The method of claim 1 , wherein the acquiring of the non-linear motion estimating function is performed by training an artificial neural network with at least the first subset of k-space data and the second subset of k-space data.

3. A computer implemented method for magnetic resonance imaging, the method comprising:

receiving a first subset of k-space data and a second subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

sampling the first subset of k-space data and the second subset of k-space data;

preprocessing the sampled first subset of k-space data and the second subset of k-space data to reduce a data dimension and/or standardize a data range of the sampled first subset of k-space data and the second subset of k-space data;

choosing the sampled first subset of k-space data as a base subset of k-space data;

estimating motion parameters of the second subset of k-space data against the base subset of k-space data; and

correcting the second subset of k-space data based on the motion parameters of the second subset of k-space data, wherein the motion parameters of the second subset of k-space data are parameters of a non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and receiving the second subset of k-space data.

4. A computer implemented method for magnetic resonance imaging, the method comprising:

receiving a first subset of k-space data and a second subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

sampling the first subset of k-space data and the second subset of k-space data;

choosing the sampled first subset of k-space data as a base subset of k-space data;

estimating motion parameters of the second subset of k-space data against the base subset of k-space data;

correcting the second subset of k-space data based on the motion parameters of the second subset of k-space data, wherein the motion parameters of the second subset of k-space data are parameters of a non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and receiving the second subset of k-space data; and

adding the corrected second subset of k-space data to the base subset of k-space data to form a second base subset of k-space data.

5. The method of claim 4 , further comprising:

receiving and/or sampling a third subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of the subject;

estimating motion parameters of the third subset of k-space data against the second base subset of k-space data;

correcting the third subset of k-space data based on the estimated motion parameters of the third subset of k-space data; and

refining the motion parameters of the second subset of k-space data.

6. The method of claim 5 , further comprising:

correcting the second base subset of k-space data based on the refined motion parameters of the second subset of k-space data.

7. The method of claim 5 , further comprising:

adding the corrected third subset of k-space data to the second base subset of k-space data to form a third base subset of k-space data,

wherein the estimating of the motion parameters of the third subset of k-space data is performed for predefined times or repeatedly until a convergence criterion is met for improving a consistency of the corrected third subset of k-space data with the third base subset of k-space data.

8. The method of claim 5 , wherein one or more of the first subset of k-space data, the second subset of k-space data, and the third subset of k-space data comprise a line in a k-space, a spiral in the k-space, a radial in the k-space, a zig-zag arm in the k-space, or a combination thereof.

9. The method of claim 5 , wherein the correcting of the second subset of k-space data and the correcting of the third subset of k-space data is performed at a same time.

10. The method of claim 5 , wherein the correcting of the second subset of k-space data and/or the correcting of the third subset of k-space data comprises reversing the motion based on the motion parameters.

11. A computer implemented method for magnetic resonance imaging, the method comprising:

receiving a first subset of k-space data and a second subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

sampling the first subset of k-space data and the second subset of k-space data;

choosing the sampled first subset of k-space data as a base subset of k-space data;

estimating motion parameters of the second subset of k-space data against the base subset of k-space data;

evaluating the second subset of k-space data based on the estimated motion parameters of the second subset of k-space data and comparing the second subset of k-space data with a predefined threshold value; and

correcting the second subset of k-space data based on the motion parameters of the second subset of k-space data, wherein the motion parameters of the second subset of k-space data are parameters of a non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and receiving the second subset of k-space data; or

reacquiring the second subset of k-space data.

12. The method of claim 11 , wherein the evaluating of the second subset of k-space data is also based on a quality of an image generated by the magnetic resonance imaging based on a k-space resulting from the first subset of k-space data and the second subset of k-space data.

13. A computer implemented method for magnetic resonance imaging, the method comprising:

providing a k-space comprising raw data corresponding to radio frequency signals emitted from excited hydrogen atoms of a subject;

receiving at least a first subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of the subject;

sampling the first subset of k-space data;

choosing the k-space comprising the raw data as a base subset of k-space data;

estimating motion parameters of the first subset of k-space data against the base subset of k-space data;

acquiring a non-linear motion estimating function by function approximation by training a neural network; and

correcting the first subset of k-space data based on the motion parameters of the first subset of k-space data, wherein the motion parameters of the first subset of k-space data are parameters of the non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and the k-space.

14. The method of claim 13 , wherein the acquiring of the non-linear motion estimating function is performed by training an artificial neural network with at least the first subset of k-space data.

15. A system for magnetic resonance imaging, the system comprising:

at least a receiving coil configured to receive subsets of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

a sampling unit configured to sample the received subsets of k-space data;

a choosing unit configured to choose a base subset of k-space data from one of the sampled subsets of k-space data or a k-space comprising raw data corresponding to radio frequency signals emitted from excited hydrogen atoms of the subject;

an estimation unit configured to estimate motion parameters of the other sampled subsets of k-space data or all of the sampled subsets of k-space data against the base subset of k-space data; and

a correcting unit configured to correct the other sampled subsets of k-space data or all of the sampled subsets of k-space data based on the motion parameters of the other sampled subsets of k-space data or all of the sampled subsets of k-space data,

wherein the motion parameters of the other sampled subsets of k-space data or all of the sampled subsets of k-space data are parameters of a non-linear motion estimating function representing a motion of the subject between receiving the chosen base subset of k-space data or the k-space comprising the raw data and receiving the other subsets of k-space data or all of the subsets of k-space data, and

wherein the system is configured to acquire the non-linear motion estimating function by function approximation by training a neural network.

16. The system of claim 15 , wherein the acquisition of the non-linear motion estimating function is configured to be performed by training an artificial neural network with the received subsets of k-space data.

17. A non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to:

receive a first subset of k-space and a second subset of k-space data as radio frequency signals emitted from excited hydrogen atoms of a subject;

sample the first subset of k-space data and the second subset of k-space data;

choose the sampled first subset of k-space data as a base subset of k-space data;

estimate motion parameters of the second subset of k-space data against the base subset of k-space data;

acquire a non-linear motion estimating function by function approximation by training a neural network; and

correct the second subset of k-space data based on the motion parameters of the second subset of k-space data, wherein the motion parameters of the second subset of k-space data are parameters of the non-linear motion estimating function representing a motion of the subject between receiving the first subset of k-space data and receiving the second subset of k-space data.

18. The non-transitory computer-readable medium of claim 17 , wherein the acquisition of the non-linear motion estimating function is configured to be performed by training an artificial neural network with at least the first subset of k-space data and the second subset of k-space data.