US-12625212-B2 - Systems and methods of artifact reduction in magnetic resonance images

Abstract

A computer-implemented method of reducing artifacts in multi-channel magnetic resonance (MR) images is provided. The method includes receiving a plurality of sets of MR images acquired by a radio-frequency (RF) coil assembly having a plurality of channels. Each set of MR images includes a plurality of slices of MR images acquired by one of the plurality of channels. The method also includes estimating a plurality of sets of artifacts in the plurality of sets of MR images by inputting the plurality of sets of MR images into a neural network model. Each set of artifacts corresponds to the one of the plurality of channels. The method further includes reducing artifacts in the plurality of sets of MR images based on estimated artifacts, deriving MR images of reduced artifacts by combining the MR images of reduced artifacts, and outputting the MR images of reduced artifacts.

Inventors

• Xinzeng Wang
• Sagar Mandava
• Xucheng Zhu

Assignees

• GE Precision Healthcare LLC

Dates

Publication Date

20260512

Application Date

20230607

Claims (20)

1. 1 . A computer-implemented method of reducing artifacts in multi-channel magnetic resonance (MR) images, comprising: receiving a plurality of sets of MR images of a volume in a subject, wherein the plurality of sets of MR images are acquired by a radio-frequency (RF) coil assembly having a plurality of channels, and each set of MR images includes a plurality of slices of MR images of the volume acquired by one of the plurality of channels, wherein the one of the plurality of channels corresponds to one or more RF coils in the RF coil assembly; estimating a plurality of sets of artifacts in the plurality of sets of MR images by: inputting the plurality of sets of MR images into a neural network model, the plurality of sets of MR images arranged in an order of channels, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model; and receiving the plurality of sets of artifacts from the neural network model, the plurality of sets of artifacts arranged in the order of channels, one set of artifacts per channel and corresponding to a set of MR images acquired by the channel; wherein each set of artifacts is channel dependent due to spatial locations of the one or more RF coils; reducing artifacts in the plurality of sets of MR images based on the plurality of sets of estimated artifacts deriving MR images of reduced artifacts by combining the plurality of sets of MR images of reduced artifacts; and outputting the MR images of reduced artifacts.
2. 2 . The method of claim 1 , wherein estimating a plurality of sets of artifacts further comprises: for each set of MR images, deriving a projected MR image by: projecting the set of MR images along a slice direction, wherein the projected MR image corresponds to the one of the plurality of channels; and inputting the plurality of sets of MR images by: inputting the projected MR images into the neural network model.
3. 3 . The method of claim 2 , wherein projecting the set of MR images further comprises: deriving the projected MR image by combining the set of MR images along the slice direction as a weighted combination.
4. 4 . The method of claim 2 , wherein projecting the set of MR images further comprises: deriving the projected MR image by adding the set of MR images along the slice direction.
5. 5 . The method of claim 2 , wherein projecting the set of MR images further comprises: deriving the projected MR image using maximum intensity projection in the slice direction.
6. 6 . The method of claim 1 , wherein reducing artifacts further comprises: for each set of MR images, determining an artifact indicator in the set of MR images based on a set of estimated artifacts corresponding to the set of MR images.
7. 7 . The method of claim 6 , wherein determining an artifact indicator further comprises: determining the artifact indicator using the neural network model, wherein the neural network model is configured to output the artifact indicator.
8. 8 . The method of claim 1 , wherein: reducing artifacts further comprises: clustering the plurality of sets of MR images based on the plurality of sets of estimated artifacts; and applying weighting to the plurality of sets of MR images based on a cluster to which a set of MR images belongs; and deriving MR images further comprises: combining the plurality of sets of weighted MR images.
9. 9 . The method of claim 8 , wherein reducing artifacts further comprises: determining a number of clusters and/or the weighting based on artifact indicators, signal intensity, and/or signal intensity distribution of the MR images.
10. 10 . The method of claim 1 , wherein: reducing artifacts further comprises: clustering the plurality of sets of MR images based on the plurality of sets of estimated artifacts; and removing one or more sets of MR images in a cluster that has the highest level of artifacts.
11. 11 . The method of claim 1 , wherein reducing artifacts further comprises: applying weighting to the plurality of sets of MR images based on the plurality of sets of estimated artifacts.
12. 12 . The method of claim 1 , wherein the neural network model is configured to estimate streak artifacts, annefacts, and/or motion artifacts in the MR image inputted into the neural network model, and estimating a plurality of sets of artifacts further comprises: estimating a plurality of sets of streak artifacts, annefacts, and/or motion artifacts in the plurality of sets of MR images using the neural network model.
13. 13 . A computer-implemented method of reducing artifacts in magnetic resonance (MR) images, comprising: receiving one or more sets of MR images of a volume in a subject, wherein the one or more sets of MR images are acquired by a radio-frequency (RF) coil assembly having one or more channels, and each set of MR images includes a plurality of slices of MR images of the volume acquired by one of the one or more channels, wherein the one of the one or more channels corresponds to one or more RF coils in the RF coil assembly; estimating one or more sets of artifacts in the one or more sets of MR images by: inputting the one or more sets of MR images into a neural network model, the one or more sets of MR images arranged in an order of channels, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model; and receiving the one or more sets of artifacts from the neural network model, the one or more sets of artifacts arranged in the order of channels, one set of artifacts per channel and corresponding to a set of MR images acquired by the channel; wherein each set of artifacts is channel dependent due to spatial locations of the one or more RF coils; and outputting the one or more sets of estimated artifacts.
14. 14 . The method of claim 13 , wherein estimating one or more sets of artifacts further comprises: for each set of MR images, deriving a projected MR image by: projecting the set of MR images along a slice direction, wherein the projected MR image corresponds to the one of the one or more channels; and inputting the one or more sets of MR images by: inputting the projected MR images into the neural network model.
15. 15 . The method of claim 13 , wherein the method further comprises: for each set of MR images, determining an artifact indicator in the set of MR images based on a set of estimated artifacts corresponding to the set of MR images.
16. 16 . The method of claim 15 , wherein determining an artifact indicator further comprises: determining the artifact indicator using the neural network model, wherein the neural network model is configured to output the artifact indicator.
17. 17 . The method of claim 13 , wherein the neural network model is configured to estimate streak artifacts, annefacts, and/or motion artifacts in the MR image inputted into the neural network model, and estimating one or more sets of artifacts further comprises: estimating one or more sets of streak artifacts, annefacts, and/or motion artifacts in the one or more sets of MR images using the neural network model.
18. 18 . A computer-implemented method of reducing artifacts in multi-channel magnetic resonance (MR) images, comprising: receiving a plurality of sets of first MR images of a volume in a subject, wherein the plurality of sets of first MR images are acquired by a radio-frequency (RF) coil assembly having a plurality of channels, wherein one of the plurality of channels corresponds to one or more RF coils in the RF coil assembly; reducing artifacts in the plurality of sets of first MR images based on a plurality of sets of estimated artifacts, the plurality of sets of first MR images arranged in an order of channels wherein the plurality of sets of estimated artifacts are estimated by: inputting a plurality of sets of second MR images into a neural network model, the plurality of sets of second MR images arranged in the order of channels, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model, and the plurality of sets of second MR images are MR images of the volume acquired by the RF coil assembly; and receiving the plurality of sets of estimated artifacts from the neural network model, the plurality of sets of estimated artifacts arranged in the order of channels, one set of artifacts per channel and corresponding to a set of MR images acquired by the channel; wherein each set of artifacts is channel dependent due to spatial locations of the one or more RF coils; deriving first MR images of reduced artifacts by combining the plurality of sets of first MR images of reduced artifacts; and outputting the first MR images of reduced artifacts.
19. 19 . The method of claim 1 , wherein the neural network model is trained with generated training images including generated natural images with channel-dependent artifacts.
20. 20 . The method of claim 13 , wherein the neural network model is trained with generated training images including generated natural images with channel-dependent artifacts.

Description

BACKGROUND The field of the disclosure relates generally to systems and methods of medical imaging, and more particularly, to systems and methods of artifact reduction in magnetic resonance (MR) images. Magnetic resonance imaging (MRI) has proven useful in diagnosis of many diseases. MRI provides detailed images of soft tissues, abnormal tissues such as tumors, and other structures, which cannot be readily imaged by other imaging modalities like computed tomography (CT). Further, MRI operates without exposing patients to ionizing radiation experienced in modalities such as CT and x-rays. Artifacts in MR images affect image quality of the images and may interfere with diagnosis. Known methods are disadvantaged in some aspects and improvements are desired. BRIEF DESCRIPTION In one aspect, a computer-implemented method of reducing artifacts in multi-channel magnetic resonance (MR) images is provided. The method includes receiving a plurality of sets of MR images of a volume in a subject. The plurality of sets of MR images are acquired by a radio-frequency (RF) coil assembly having a plurality of channels, and each set of MR images includes a plurality of slices of MR images of the volume acquired by one of the plurality of channels. The method also includes estimating a plurality of sets of artifacts in the plurality of sets of MR images by inputting the plurality of sets of MR images into a neural network model, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model. Each set of artifacts corresponds to the one of the plurality of channels. The method further includes reducing artifacts in the plurality of sets of MR images based on the plurality of sets of estimated artifacts, deriving MR images of reduced artifacts by combining the plurality of sets of MR images of reduced artifacts, and outputting the MR images of reduced artifacts. In another aspect, a computer-implemented method of reducing artifacts in MR images is provided. The method includes receiving one or more sets of MR images of a volume in a subject. The one or more sets of MR images are acquired by an RF coil assembly having one or more channels, and each set of MR images includes a plurality of slices of MR images of the volume acquired by one of the one or more channels. The method further includes estimating one or more sets of artifacts in the one or more sets of MR images by inputting the one or more sets of MR images into a neural network model, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model. Each set of artifacts corresponds to the one of the one or more channels. The method further includes outputting the one or more sets of estimated artifacts. In one more aspect, a computer-implemented method of reducing artifacts in multi-channel MR images is provided. The method includes receiving a plurality of sets of first MR images of a volume in a subject, wherein the plurality of sets of first MR images are acquired by an RF coil assembly having a plurality of channels. The method also includes reducing artifacts in the plurality of sets of first MR images based on a plurality of sets of estimated artifacts. The plurality of sets of estimated artifacts are estimated by inputting a plurality of sets of second MR images into a neural network model, wherein the neural network model is configured to estimate artifacts in an MR image inputted into the neural network model. The plurality of sets of second MR images are MR images of the volume acquired by the RF coil assembly. Each set of artifacts corresponds to the one of the plurality of channels. The method further includes deriving first MR images of reduced artifacts by combining the plurality of sets of first MR images of reduced artifacts, and outputting the first MR images of reduced artifacts. DRAWINGS FIG. 1 is a schematic diagram of an example magnetic resonance imaging (MRI) system. FIG. 2 is a schematic diagram showing combination of images acquired by channels of a radio-frequency (RF) coil assembly. FIG. 3A is an example artifact reduction system. FIG. 3B is a flow chart of an example method of reducing artifacts. FIG. 4 is a schematic diagram illustrating the work flow of an example embodiment of the method shown in FIG. 3B. FIG. 5 is a schematic diagram showing that a neural network model is used to estimate artifacts. FIG. 6A is a plot of artifact indicators of the channels having two clusters. FIG. 6B is a plot of artifact indicators of the channels having three clusters. FIG. 7 is a comparison of images reconstructed with and without using the systems and methods described herein. FIG. 8A is a schematic diagram of a neural network model. FIG. 8B is a schematic diagram of a neuron in the neural network model shown in FIG. 8A. FIG. 9 is a block diagram of an example computing device. FIG. 10 is a block diagram of an example server computin