Method and device for water-fat separation of image, computer apparatus, and computer-readable storage medium
View Patent ↗Method and device for water-fat separation of image, computer apparatus, and computer-readable storage medium. The method includes: obtaining plurality of echo images of a target object; obtaining an initial water and fat distribution image of the target object; determining a first error phasor candidate and a second error phasor candidate, the first error phasor candidate being used to characterize an error phasor of each element corresponding to water occupying a major constituent, and the second error phasor candidate being used to characterize an error phasor of each element corresponding to fat occupying a major constituent; determining an initial-guess of error phasor corresponding to each element in the plurality of echo images among the first error phasor candidate and the second error phasor candidate based on the initial water and fat distribution image; determining an optimal error phasor based on the initial-guess of the error phasor; and acquiring a water image and a fat image of the target object from the plurality of echo images according to the optimal error phasor.
1 . A method for water-fat separation of image, comprising:
obtaining a plurality of echo images of a target object;
obtaining an initial water and fat distribution image of the target object;
determining a first error phasor candidate and a second error phasor candidate, the first error phasor candidate being used to represent an error phasor of each element corresponding to water occupying a major constituent in the element, and the second error phasor candidate being used to represent an error phasor of each element corresponding to fat occupying a major constituent in the element;
determining an initial-guess of error phasor corresponding to each element in the plurality of echo images among the first error phasor candidate and the second error phasor candidate based on the initial water and fat distribution image; and
determining an optimal error phasor based on the initial-guess of the error phasor, and acquiring a water image and a fat image of the target object from the plurality of echo images according to the optimal error phasor.
2 . The method for water-fat separation of image of claim 1 , wherein the acquiring the initial water and fat distribution image of the target object comprises:
performing a tissue segmentation on the plurality of echo images to obtain a plurality of tissue-segmented images; and
determining water-fat ratios of each element in the plurality of tissue-segmented images according to prior knowledge of water-fat ratios of various tissues, to obtain the initial water and fat distribution image.
3 . The method for water-fat separation of image of claim 1 , wherein the acquiring the initial water and fat distribution image of the target object comprises:
obtaining fat-suppressed images of the target object; and
comparing the plurality of echo images with the fat-suppressed images to obtain the initial water and fat distribution image.
4 . The method for water-fat separation of image of claim 1 , wherein the determining the first error phasor candidate and the second error phasor candidate comprises:
determining a first amplitude candidate and a second amplitude candidate of water and fat signals of the target object based on the plurality of echo images and a predefined water-fat signal model; and
calculating the first error phasor candidate and the second error phasor candidate according to the first amplitude candidate, the second amplitude candidate, and the predefined water-fat signal model.
5 . The method for water-fat separation of image of claim 1 , wherein the determining the first error phasor candidate and the second error phasor candidate comprises:
determining the first error phasor candidate and the second error phasor candidate by using a variable projection algorithm and a maximum likelihood estimation, the plurality of echo images comprising at least three echo images.
6 . The method for water-fat separation of image of claim 1 , wherein the determining the first error phasor candidate and the second error phasor candidate comprises:
determining the first error phasor candidate and the second error phasor candidate by solving a triangle, the plurality of echo images comprising two echo images.
7 . The method for water-fat separation of image of claim 4 , wherein the plurality of echo images comprise an in-phase image, and a partially-opposed-phase image.
8 . The method for water-fat separation of image of claim 7 , wherein the first amplitude candidate and the second amplitude candidate of the water and fat signals of the target object are determined based on the plurality of echo images and the predefined water-fat signal model to be:
B
=
1
2
❘
"\[LeftBracketingBar]"
M
1
+
2
M
2
2
-
M
1
2
(
1
+
cos
α
)
1
-
cos
α
❘
"\[RightBracketingBar]"
S
=
1
2
❘
"\[LeftBracketingBar]"
M
1
-
2
M
2
2
-
M
1
2
(
1
+
cos
α
)
1
-
cos
α
❘
"\[RightBracketingBar]"
wherein B denotes the first amplitude candidate, S denotes the second amplitude candidate, M 1 denotes a signal amplitude of the in-phase image, M 2 denotes a signal amplitude of the partially-opposed-phase image, and a denotes an phase angle between water and fat in the partially-opposed-phase image.
9 . The method for water-fat separation of image of claim 1 , wherein the determining the initial-guess of error phasor corresponding to each element in the plurality of echo images among the first error phasor candidate and the second error phasor candidate based on the initial water and fat distribution image, comprises:
for each element in the initial water and fat distribution image,
using the first error phasor candidate as the initial-guess of error phasor in response to the element indicating a signal of water occupying the major constituent; and
using the second error phasor candidate as the initial-guess of error phasor in response to the element indicating a signal of fat occupying the major constituent.
10 . The method for water-fat separation of image of claim 2 , wherein the determining the first error phasor candidate and the second error phasor candidate comprises:
determining the first amplitude candidate and the second amplitude candidate of water and fat signals of the target object based on the plurality of echo images and a predefined water-fat signal model; and
calculating the first error phasor candidate and the second error phasor candidate according to the first amplitude candidate, the second amplitude candidate and the predefined water-fat signal model.
11 . The method for water-fat separation of image of claim 3 , wherein the determining the first error phasor candidate and the second error phasor candidate comprises:
determining the first amplitude candidate and the second amplitude candidate of water and fat signals of the target object based on the plurality of echo images and a predefined water-fat signal model; and
calculating the first error phasor candidate and the second error phasor candidate according to the first amplitude candidate, the second amplitude candidate and the predefined water-fat signal model.
12 . The method for water-fat separation of image of claim 1 , wherein the determining the optimal error phasor from the initial-guess of the error phasor, and acquiring the water image and the fat image of the target object from the plurality of echo images according to the optimal error phasor, comprise:
performing an iterative smooth filtering processing on the initial-guess of error phasor to obtain the optimal error phasor;
obtaining a signal model matrix equation by using the optimal error phasor; and
solving the signal model matrix equation by a least square method to obtain the water image and the fat image of the target object.
13 . The method for water-fat separation of image of claim 1 , wherein at least one element of the initial water and fat distribution image comprises a fat fraction.
14 . The method for water-fat separation of image of claim 1 , wherein the plurality of echo images are two-dimensional images, and the element in each of the plurality of echo images is a pixel.
15 . The method for water-fat separation of image of claim 1 , wherein the plurality of echo images are three-dimensional images, and the element in each of the plurality of echo images is a voxel.
16 . A computer product, comprising a non-transitory computer readable medium storing a computer program, wherein, the computer program, when executed by a processor, causes the processor to perform steps of the method of claim 1 .