Systems and methods for magnetic resonance imaging

• US 10,638,951 B2
• Filed: 12/24/2018
• Issued: 05/05/2020
• Est. Priority Date: 07/06/2018
• Status: Active Grant

• Current Assignee: Shanghai United Imaging Healthcare Co., Ltd.
• Original Assignee: Shanghai United Imaging Healthcare Co., Ltd.
• Inventors: Ye, Yongquan, Zong, Jinguang
• Primary Examiner(s): Fuller, Rodney E
• Application Number:
  US16/231,721
• Publication Number:
  US 20200008701A1
• Time in Patent Office: 498 Days
• Field of Search: 324307
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: G01R 33/243 Spatial mapping of the polarizing magnetic field
• CPC Class Codes: G01R 33/385 using gradient magnetic field coils
• CPC Class Codes: G01R 33/443 Assessment of an electric or a magnetic field, e.g. spatial mapping, determination of a B0 drift or dosimetry
• CPC Class Codes: G01R 33/54 Signal processing systems, e.g. using pulse sequences ; Generation or control of pulse sequences; Operator console
• CPC Class Codes: G01R 33/56 Image enhancement or correction, e.g. subtraction or averaging techniques , e.g. improvement of signal-to-noise ratio and resolution
• CPC Class Codes: G01R 33/5608 Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels image data processing or generation, in general G06T

Abstract

A method may include obtaining first image data of a first subject and obtaining a target machine learning model. The first image data may indicate a first intensity distribution of a first magnetic field relative to at least one portion of the first subject. The target machine learning model may provide a mapping relationship between second intensity distributions of second magnetic fields and corresponding second susceptibility distributions associated with second subjects under the second magnetic fields. The method may also include generating a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data. The method may further include determining a target MR image of the at least one portion of the first subject based on the first susceptibility distribution.

Claims

1. A system for magnetic resonance imaging (MRI), comprising:

at least one storage device storing executable instructions, and

at least one processor in communication with the at least one storage device, when executing the executable instructions, causing the system to;

obtain first image data of a first subject, the first image data indicating a first intensity distribution of a first magnetic field relative to at least one portion of the first subject;

obtain a target machine learning model that provides a mapping relationship between second intensity distributions of second magnetic fields and corresponding second susceptibility distributions associated with second subjects under the second magnetic fields;

generate a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject; and

determine a target MR image of the at least one portion of the first subject based on the first susceptibility distribution associated with the first subject.

2. The system of claim 1, wherein the first image data includes at least one of an MR complex diagram, an MR phase diagram, or an MR field diagram.

3. The system of claim 1, wherein to generate a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject, the at least one processor is further configured to cause the system to:

input the first image data of the first subject into the target machine learning model; and

obtain the first susceptibility distribution associated with the first subject outputted by the target machine learning model based on the mapping relationship.

4. The system of claim 1, wherein to generate a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject, the at least one processor is further configured to cause the system to:

identify the first intensity distribution of the first magnetic field from the first image data;

input the first intensity distribution of the first magnetic field into the target machine learning model; and

obtain the first susceptibility distribution associated with the first subject outputted by the target machine learning model based on the mapping relationship.

5. The system of claim 1, wherein the at least one processor is further configured to cause the system to:

perform a domain transformation operation on at least one of the first image data of the first subject or the first susceptibility distribution associated with the first subject.

6. The system of claim 5, wherein the at least one processor is further configured to cause the system to:

perform a first domain transformation operation on the first image data of the first subject; and

perform a second domain transformation operation on the first susceptibility distribution associated with the first subject, the second domain transformation operation being a reverse operation of the first domain transformation operation.

7. The system of claim 1, wherein the at least one processor is further configured to cause the system to:

generate the first susceptibility distribution associated with the first subject based on the mapping relationship and at least one of a scanning parameter associated with the first image data, a dipole kernel function associated with the first image data, a feature relating to a reference subject around the first subject, a feature relating to the at least one portion of the first subject, or a spatial relationship between the first subject and the reference subject around the first subject.

8. The system of claim 1, wherein to obtain a target machine learning model, the at least one processor is further configured to cause the system to:

obtain a plurality of first groups of training data, each group of the plurality of first groups of training data including second image data and a first reference susceptibility distribution associated with a third subject, the second image data indicating a third intensity distribution of a third magnetic field relative to the third subject;

train a machine learning model based on the plurality of first groups of training data to obtain a first machine learning model; and

determine the target machine learning model based on the first machine learning model.

9. The system of claim 8, wherein to train a machine learning model based on the plurality of first groups of training data, the at least one processor is further configured to cause the system to:

for each group of the plurality of first groups of training data, identify the third intensity distribution of the third magnetic field from the second image data; and

train the machine learning model using the identified third intensity distribution of the third magnetic field and the first reference susceptibility distribution associated with the third subject corresponding to the each group of the plurality of first groups of training data.

10. The system of claim 8, wherein to determine the target machine learning model based on the first machine learning model, the at least one processor is further configured to cause the system to:

obtain a plurality of second groups of training data, each group of the plurality of groups of training data including third image data and a second reference susceptibility distribution associated with the first subject;

train the first machine learning model using the plurality of second groups of training data to obtain a second machine learning model; and

designate the second machine learning model as the target machine leaning model.

11. The system of claim 8, wherein to obtain a plurality of first groups of training data, the at least one processor is further configured to cause the system to:

for each of one or more groups of the plurality of first groups of training data,

determine one or more first reference susceptibility distributions associated with the third subject;

simulate the second image data based on the corresponding first reference susceptibility distributions; and

determine the one or more groups of the plurality of first groups of training data based on the one or more first reference susceptibility distributions and the simulated second image data.

12. The system of claim 8, wherein to train a machine learning model based on the plurality of first groups of training data, the at least one processor is further configured to cause the system to:

obtain an objective function including a regularization configured to constrain the training of the machine learning model to converge the objective function; and

perform a plurality of iterations based on the objective function using the plurality of first groups of training data.

13. The system of claim 12, wherein the regularization is constructed based on at least one of a scanning parameter associated with the second image data, a dipole kernel function associated with the second image data, a feature relating to a reference subject around the third subject, a feature relating to the at least one portion of the third subject, or a spatial relationship between the third subject and the reference subject around the third subject.

14. A method implemented on a computing device including a storage device and at least one processor for magnetic resonance imaging, the method comprising:

obtaining first image data of a first subject, the first image data indicating a first intensity distribution of a first magnetic field relative to at least one portion of the first subject;

obtaining a target machine learning model that provides a mapping relationship between second intensity distributions of second magnetic fields and corresponding second susceptibility distributions associated with second subjects under the second magnetic fields;

generating a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject; and

determining a target MR image of the at least one portion of the first subject based on the first susceptibility distribution associated with the first subject.

15. The method of claim 14, wherein the generating, based on the target machine learning model and the first image data of the first subject, a first susceptibility distribution associated with the first subject under the first magnetic field comprising:

inputting the first image data of the first subject into the target machine learning model; and

obtaining the first susceptibility distribution associated with the first subject outputted by the target machine learning model based on the mapping relationship.

16. The method of claim 14, wherein the generating a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject comprising:

identifying the first intensity distribution of the first magnetic field from the first image data;

inputting the first intensity distribution of the first magnetic field into the target machine learning model; and

obtaining the first susceptibility distribution associated with the first subject outputted by the target machine learning model based on the mapping relationship.

17. The method of claim 14, further comprise:

performing a domain transformation operation on at least one of the first image data of the first subject or the first susceptibility distribution associated with the first subject.

18. The method of claim 14, further comprise:

generating the first susceptibility distribution associated with the first subject based on the mapping relationship and at least one of a scanning parameter associated with the first image data, a dipole kernel function associated with the first image data, a feature relating to a reference subject around the first subject, a feature relating to the at least one portion of the first subject, or a spatial relationship between the first subject and the reference subject around the first subject.

19. The method of claim 14, wherein the obtaining the target machine learning model comprising:

obtaining a plurality of first groups of training data, each group of the plurality of first groups of training data including second image data and a first reference susceptibility distribution associated with a third subject, the second image data indicating a third intensity distribution of a third magnetic field relative to the third subject;

training a machine learning model based on the plurality of first groups of training data to obtain a first machine learning model; and

determining the target machine learning model based on the first machine learning model.

20. A non-transitory computer readable medium storing instructions, the instructions, when executed by at least one processor, causing the at least one processor to implement a method comprising:

obtaining first image data of a first subject, the first image data indicating a first intensity distribution of a first magnetic field relative to at least one portion of the first subject;

obtaining a target machine learning model that provides a mapping relationship between second intensity distributions of second magnetic fields and corresponding second susceptibility distributions associated with second subjects under the second magnetic fields;

generating a first susceptibility distribution associated with the first subject under the first magnetic field based on the target machine learning model and the first image data of the first subject; and

determining a target MR image of the at least one portion of the first subject based on the first susceptibility distribution associated with the first subject.