System and method for generating attenuation map

1. A method of generating an attenuation map for positron emission tomography (PET) reconstruction, the method comprising:

acquiring PET data indicative of a subject;

acquiring an anatomic image indicative of the subject, wherein the anatomic image comprises a plurality of voxels;

fetching a reference image from a database, wherein the reference image comprises voxel segmentation information;

registering the reference image to the anatomic image;

segmenting the anatomic image into a plurality of regions based on the voxel segmentation information of the reference image;

generating a first attenuation map corresponding to the anatomic image by assigning attenuation coefficients to the plurality of regions of the anatomic image;

for a voxel of the plurality of voxels of the anatomic image,

determining a three-dimensional (3D) region in the anatomic image, wherein the voxel is the center of the 3D region;

calculating a registration accuracy of the 3D region between the anatomic image and the reference image;

determining a probability distribution of attenuation coefficient of the voxel based on the registration accuracy; and

iteratively updating the first attenuation map based on the probability distribution of attenuation coefficient of the voxel and the PET data to obtain a final attenuation map.

2. The method of claim 1 , the fetching the reference image is based on one or more types of characteristic information of the subject.

3. The method of claim 2 , wherein the one or more types of characteristic information comprise height, weight, gender of the subject, or an area of the subject to imaging.

4. The method of claim 1 , the registering the reference image to the anatomic image is based on at least one of optical flow, feature points, a contour, or grey scale information.

5. The method of claim 1 , the determining the probability distribution of attenuation coefficient of the voxel comprising:

calculating a statistical probability indicative of the voxel belonging to a certain region of the plurality of regions based on the registration accuracy; and

acquiring the probability distribution of attenuation coefficient of the voxel based on the statistical probability.

6. The method of claim 5 , wherein the at least one region of the plurality of regions is a spatial region.

7. The method of claim 5 further comprising

designating, if the statistical probability is 1, the voxel as belonging to the certain region of the plurality of regions.

8. The method of claim 7 , wherein the certain region of the plurality of regions comprises a plurality of voxels with a same probability distribution of attenuation coefficient, and the plurality of voxels of the certain region are assigned with a same attenuation coefficient.

9. The method of claim 5 further comprising

designating, if the statistical probability is less than 1, the voxel as belonging to at least two regions of the plurality of regions.

10. The method of claim 9 , the voxel is designated as belonging to one region of the at least two regions as the final attenuation map is obtained.

11. The method of claim 1 , wherein the anatomic image is a magnetic resonance (MR) image or a computed tomography (CT) image.

12. A method of image reconstruction for positron emission tomography (PET) comprising:

acquiring PET data indicative of a subject;

acquiring a PET image by reconstructing the PET data, wherein the PET data is attenuation corrected by an attenuation map, wherein the attenuation map is generated by following steps:

acquiring an MR image indicative of the subject, wherein the MR image comprises a plurality of voxels;

fetching a reference image from a database, wherein the reference image comprises voxel segmentation information;

registering the reference image to the MR image;

segmenting the MR image into a plurality of regions based on the voxel segmentation information of the reference image;

generating a first attenuation map corresponding to the MR image by assigning attenuation coefficients stored in the database to the plurality of regions of the MR image;

for a voxel of the plurality of voxels of the MR image,

determining a three-dimensional (3D) region in the MR image, wherein the voxel is the center of the 3D region;

calculating a registration accuracy of the 3D region between the MR image and the reference image;

determining a probability distribution of attenuation coefficient of the voxel based on the registration accuracy;

updating the PET data based on the first attenuation map and generating a first PET image; and

updating the first attenuation map based on the probability distribution of attenuation coefficient and the PET data of the first PET image to obtain a second attenuation map.

13. The method of claim 12 , the updating the first attenuation map based on the probability distribution of attenuation coefficient and the PET data of the first PET image to obtain the second attenuation map comprising:

acquiring the probability distribution of attenuation coefficient of the voxel in one or more regions based on the registration accuracy; and

acquiring the second attenuation map based on a function relating to the probability distribution and the PET data of the first PET image.

14. The method of claim 13 , wherein the function is a penalized likelihood function.

15. The method of claim 12 , further comprising:

updating the PET data of the first PET image to generate a second PET image while keeping the second attenuation map fixed; and

updating the second attenuation map to generate a third attenuation map while keeping the PET data of the second PET image fixed.

16. The method of claim 12 , the segmenting the MR image into a plurality of regions based on the voxel segmentation information of the reference image comprising:

assigning the voxels belonging to one region of the plurality of regions with an attenuation coefficient.

17. An imaging system comprising:

a storage configured to store PET data of a subject and a set of instructions;

at least one processor in communication with the storage, wherein when executing the instructions, the at least one processor is configured to cause the imaging system to:

receive an anatomic image indicative of the subject, wherein the anatomic image comprises a plurality of voxels;

fetch a reference image from a database, wherein the reference image comprises voxel segmentation information;

register the reference image to the anatomic image;

segment the anatomic image into a plurality of regions based on the voxel segmentation information;

generate a first attenuation map corresponding to the anatomic image by assigning attenuation coefficients to the plurality of regions of the anatomic image;

for a voxel of the plurality of voxels of the anatomic image,

determining a three-dimensional (3D) region in the anatomic image, wherein the voxel is the center of the 3D region;

calculate a registration accuracy of the 3D region between the anatomic image and the reference image;

determining a probability distribution of attenuation coefficient of the voxel based on the registration accuracy;

update the PET data based on the first attenuation map and generating a first PET image;

update the first attenuation map based on the probability distribution of attenuation coefficient and the PET data of the first PET image to obtain a second attenuation map.

18. The imaging system of claim 17 , wherein the anatomic image is a magnetic resonance (MR) image or a computed tomography (CT) image.

19. The imaging system of claim 17 , wherein the at least one processor is configured to cause the imaging system further to:

receive characteristic information of the subject.

20. The imaging system of claim 19 , wherein the fetching the reference image is based on the characteristic information of the subject.