System and method for reducing artifacts in echo planar magnetic resonance imaging

1. A method for reducing N/2 ghost or Nyquist ghost in magnetic resonance (MR) images, the method comprising:

acquiring k-space dataset for an object using an echo planar imaging (EPI) sequence;

dividing the k-space dataset into first partial k-space subset data related to positive echoes and second partial k-space subset data related to negative echoes;

obtaining third partial k-space subset data that is N/2 or Nyquist ghost-free subset data;

respectively registering the first partial k-space subset data and the second partial k-space subset data to a first portion of the third partial k-space subset data corresponding to positive echoes and a second portion of the third partial k-space subset data corresponding to negative echoes;

combining the registered first partial k-space subset data and the registered second partial k-space subset data to form full k-space dataset; and

reconstructing an image for the object based on the full k-space dataset.

2. The method of claim 1 , wherein the echo planar imaging sequence is either single-shot EPI sequence or multi-shots EPI sequence.

3. The method of claim 1 , wherein the third partial k-space subset data is acquired by the EPI sequence at a time frame different from a time frame when the first partial k-space subset data and the second partial k-space subset data are acquired.

4. The method of claim 1 wherein the third partial k-space subset data is acquired by another EPI sequence different from the EPI sequence, and the third partial k-space subset data and the first and second partial k-space subset data are acquired at a same modality or different modalities.

5. The method of claim 1 , wherein acquiring the third partial k-space subset data comprise:

obtaining the third partial k-space subset data from the acquired k-space dataset; or

obtaining the third partial k-space subset data from additional MR scanning.

6. The method of claim 5 , wherein the k-space dataset is acquired using non-uniform phase-encoding; and

the third partial k-space subset data is obtained from the k-space dataset acquired using the non-uniform phase encoding.

7. The method of claim 1 , wherein registering the first partial k-space subset data and the second partial k-space subset data to the third partial k-space subset data comprises:

interpolating the third partial k-space subset data into increased resolution k-space dataset which resolution is equal to resolution of the acquired k-space dataset;

estimating a first transformation matrix based on the first partial k-space subset data and the increased resolution k-space dataset corresponding to the first partial k-space subset data;

estimating a second transformation matrix based on the second partial k-space subset data and the increased resolution k-space dataset corresponding to the second partial k-space subset data;

correcting the first partial k-space subset data using the first transformation matrix; and

correcting the second partial k-space subset data using the second transformation matrix.

8. The method of claim 7 , wherein the first transformation matrix reflects variations of translation, rotation or shear data between the first partial k-space subset data and increased resolution k-space subset data, and

the second transformation matrix reflects variations of translation, rotation or shear data between the second partial k-space subset data and the increased resolution k-space subset data.

9. The method of claim 1 , wherein combining the registered first partial k-space subset data and the registered second partial k-space subset data to form the full k-space dataset comprises:

filling the registered first partial k-space subset data to positive echoes of the full k-space dataset; and

filling the registered second partial k-space subset data to negative echoes of the full k-space dataset.

10. A method for reducing N/2 ghost or Nyquist ghost in magnetic resonance (MR) images, the method comprising:

acquiring k-space dataset for an object using an echo planar imaging (EPI) sequence;

dividing the k-space dataset into first partial k-space subset data related to even echoes and second partial k-space subset data related to odd echoes;

obtaining third partial k-space subset data that is N/2 or Nyquist ghost-free subset data;

estimating relational information between each even k-space data and odd k-space data adjacent to the each odd k-space data based on the ghost-free reference k-space dataset;

estimating another relational information between each odd k-space data and even k-space data adjacent to the each even k-space data based on the ghost-free reference k-space dataset;

synthesizing missing k-space lines related to odd echoes in the first partial k-space subset data based on the first partial k-space subset data and the relational information;

synthesizing missing k-space lines related to even echoes in the second partial k-space subset data based on the second partial k-space subset data and the another relational information;

creating a first full k-space dataset related to both odd and even echoes by combining the first partial k-space subset data and the synthesized k-space lines related to odd echoes;

creating a second full k-space dataset related to both odd and even echoes by combining the second partial k-space subset data and the synthesized k-space lines related to even echoes;

reconstructing a first ghost-free EPI image from the first full k-space dataset;

reconstructing a second ghost-free EPI images from the second full k-space dataset; and

creating a ghost-free EPI image using the first and second ghost-free EPI images.

11. The method of claim 10 , wherein obtaining third partial k-space subset data that is N/2 or Nyquist ghost-free subset data further comprises:

obtaining third partial k-space subset data that is N/2 or Nyquist ghost-free subset data corresponding to the first partial k-space subset data; and

obtaining third partial k-space subset data that is N/2 or Nyquist ghost-free subset data corresponding to the second partial k-space subset data.

12. The method of claim 10 , wherein the relational information includes relational information between each of the missing k-space lines of the first partial k-space subset data and each of k-space lines that are adjacent to the missing k-space lines in the third partial k-space subset data corresponding to the first partial k-space subset data.

13. The method of claim 10 , wherein the additional relational information includes relational information between each of the missing k-space lines of the second partial k-space subset data and each of k-space lines that are adjacent to the missing k-space lines in the third partial k-space subset data corresponding to the second partial k-space subset data.

14. A magnetic resonance imaging (MRI) system comprising:

a magnetic field generating unit configured to apply a plurality of RF pulses with a variable flip angle to a target area in the object;

a receiver configured to receive MR signals from the target area;

a processing unit;

a system memory; and

machine readable instructions stored in the system memory that, when executed by the processing unit, cause the processing unit to:

acquire k-space dataset for an object using an echo planar imaging (EPI) sequence;

divide the k-space dataset into first partial k-space subset data related to positive echoes and second partial k-space subset data related to negative echoes;

obtain third partial k-space subset data that is N/2 or Nyquist ghost-free subset data;

respectively register the first partial k-space subset data and the second partial k-space subset data to a first portion of the third partial k-space subset data corresponding to positive echoes and a second portion of the third partial k-space subset data corresponding to negative echoes;

combine the registered first partial k-space subset data and the registered second partial k-space subset data to form full k-space dataset; and

reconstruct an image for the object based on the full k-space dataset.

15. The system of claim 14 , wherein the echo planar imaging sequence is either single-shot EPI sequence or multi-shots EPI sequence.

16. The system of claim 14 , wherein the third partial k-space subset data is acquired by the EPI sequence at a time frame different from a time frame when the first partial k-space subset data and the second partial k-space subset data are acquired.

17. The system of claim 14 , wherein the third partial k-space subset data is acquired by another EPI sequence different from the EPI sequence, and the third partial k-space subset data and the first and second partial k-space subset data are acquired at a same modality or different modalities.

18. The system of claim 14 , wherein the third partial k-space subset data is acquired from the acquired k-space dataset or from additional MR scanning.

19. The system of claim 18 , wherein the third partial k-space subset data from the acquired k-space dataset is acquired using the non-uniform phase encoding.

20. The system of claim 14 , wherein registering the first partial k-space subset data and the second partial k-space subset data to the third partial k-space subset data comprises:

interpolating the third partial k-space subset data into increased resolution k-space dataset which resolution is equal to resolution of the acquired k-space dataset;

estimating a first transformation matrix based on the first partial k-space subset data and the increased resolution k-space dataset corresponding to the first partial k-space subset data;

estimating a second transformation matrix based on the second partial k-space subset data and the increased resolution k-space dataset corresponding to the second partial k-space subset data;

correcting the first partial k-space subset data using the first transformation matrix; and

correcting the second partial k-space subset data using the second transformation matrix.