Systems and methods for attenuation correction

1. A system for attenuation correction 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 at least one first positron emission tomography (PET) image of a subject acquired by a PET scanner;

obtain at least one first magnetic resonance (MR) image of the subject acquired by an MR scanner;

obtain a target neural network model that provides a mapping relationship between at least one second PET image acquired by the PET scanner, at least one second MR image acquired by the MR scanner, and a second attenuation correction data;

express the at least one first PET image of the subject acquired by the PET scanner in a form of a first matrix;

generate, based on the mapping relationship between the second attenuation correction data, the at least one second PET image, and the at least one second MR image, a first attenuation correction data in a form of a second matrix;

and

determine, based on the second matrix, a target PET image of the subject by correcting the first matrix.

2. The system of claim 1 , wherein to obtain a target neural network model, the at least one processor is further configured to cause the system to:

obtain multiple groups of training data, each of the multiple groups of training data including the at least one second PET image, the at least one second MR image, and the second attenuation correction data corresponding to a sample; and

generate the target neural network model by training a neural network model using the multiple groups of training data.

3. The system of claim 2 , wherein the neural network model includes at least one of a convolutional neural network (CNN) model, a back propagation (BP) neural network model, a radial basis function (RBF) neural network model, a deep belief nets (DBN) neural network model, or an Elman neural network model.

4. The system of claim 2 , wherein to obtain multiple groups of training data, the at least one processor is further configured to cause the system to:

for each of the multiple groups of training data,

obtain a CT image of the sample; and

determine, the second attenuation correction data corresponding to the sample based on the CT image of the sample.

5. The system of claim 2 , wherein to obtain multiple groups of training data, the at least one processor is further configured to cause the system to:

for each of the multiple groups of training data, determine, based on the at least one second MR image or the at least one second PET image, the second attenuation correction data corresponding to the sample.

6. The system of claim 1 , wherein to generate the first attenuation correction data, the at least one processor is further configured to cause the system to:

input the at least one first PET image and the at least one first MR image to the target neural network model; and

obtain the first attenuation correction data output by the target neural network model.

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

obtain PET projection data associated with the first PET image of the subject; and

reconstruct, based on the PET projection data and the first attenuation correction data, the target PET image.

8. The system of claim 7 , wherein to determine the target PET image of the subject, the at least one processor is further configured to cause the system to:

perform a post-processing operation on the first attenuation correction data corresponding to the subject, the post-processing operation including at least one of an interpolation operation or a registration operation.

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

perform a pre-processing operation on at least one of the at least one first PET image or the at least one first MR image, the pre-processing operation including at least one of a filtering operation, a smoothing operation, a transformation operation, or a denoising operation.

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

update the target neural network model based on a testing performed on the target neural network model.

11. A method for attenuation correction implemented by a computing device, the computing device including at least one processor and at least one storage device, the method comprising:

obtaining at least one first PET image of a subject acquired by a PET scanner;

obtaining at least one first MR image of the subject acquired by an MR scanner;

obtaining a target neural network model that provides a mapping relationship between at least one second PET image acquired by the PET scanner, at least one second MR image acquired by the MR scanner, and a second attenuation correction data;

expressing the at least one first PET image of the subject acquired by the PET scanner in a form of a first matrix;

generating, based on the mapping relationship between the second attenuation correction data, the at least one second PET image, and the at least one second MR image, a first attenuation correction data in a form of a second matrix;

and

determining, based on the second matrix, a target PET image of the subject by correcting the first matrix.

12. The method of claim 11 , wherein the obtaining a target neural network model includes:

obtaining multiple groups of training data, each of the multiple groups of training data including the at least one second PET image, the at least one second MR image, and the second attenuation correction data corresponding to a sample; and

generating the target neural network model by training a neural network model using the multiple groups of training data.

13. The method of claim 12 , wherein the neural network model includes at least one of a convolutional neural network (CNN) model, a back propagation (BP) neural network model, a radial basis function (RBF) neural network model, a deep belief nets (DBN) neural network model, or an Elman neural network model.

14. The method of claim 12 , wherein the obtaining multiple groups of training data includes:

for each of the multiple groups of training data,

obtaining a CT image of the sample; and

determining the second attenuation correction data corresponding to the sample based on the CT image of the sample.

15. The method of claim 12 , wherein the obtaining multiple groups of training data includes:

for each of the multiple groups of training data, determining, based on the at least one second MR image or the at least one second PET image, the second attenuation correction data corresponding to the sample.

16. The method of claim 11 , wherein the generating the first attenuation correction data includes:

inputting the at least one first PET image and the at least one first MR image to the target neural network model; and

obtaining the first attenuation correction data output by the target neural network model.

17. The method of claim 11 , wherein the determining the target PET image of the subject includes:

obtaining PET projection data associated with the first PET image of the subject; and

reconstructing, based on the PET projection data and the first attenuation correction data, the target PET image.

18. The method of claim 17 , wherein the determining the target PET image of the subject includes:

performing a post-processing operation on the first attenuation correction data corresponding to the subject, the post-processing operation including at least one of an interpolation operation or a registration operation.

19. The method of claim 11 , further comprising:

performing a pre-processing operation on at least one of the at least one first PET image or the at least one first MR image, the pre-processing operation including at least one of a filtering operation, a smoothing operation, a transformation operation, or a denoising operation.

20. A non-transitory computer-readable medium storing at least one set of instructions, wherein when executed by at least one processor, the at least one set of instructions directs the at least one processor to perform acts of:

obtaining at least one first PET image of a subject acquired by a PET scanner;

obtaining at least one first MR image of the subject acquired by an MR scanner;

obtaining a target neural network model that provides a mapping relationship between at least one second PET image acquired by the PET scanner, at least one second MR image acquired by the MR scanner, and a second attenuation correction data;

expressing the at least one first PET image of the subject acquired by the PET scanner in a form of a first matrix;

generating, based on the mapping relationship between the second attenuation correction data, the at least one second PET image and the at least one second MR image, a first attenuation correction data in a form of a second matrix;

and

determining, based on the second matrix, a target PET image of the subject by correcting the first matrix.