Reconstructing PET image

1. A method of reconstructing a Positron Emission Computed Tomography (PET) image, comprising:

determining a plurality of Lines of Response (LORs) associated with multiple-coincidence data in coincidence data detected by a PET device through scanning, wherein the coincidence data comprises double-coincidence data indicating that two single events satisfy a coincidence condition and the multiple-coincidence data indicating that three or more single events satisfy the coincidence condition, and each of the LORs is a line between crystals corresponding to every two of the three or more single events corresponding to the multiple-coincidence data;

obtaining, for each of the LORs, a respective line integral value along the LOR according to a time difference between two single events corresponding to the LOR;

allocating the multiple-coincidence data to the LORs according to the obtained respective line integral values of the LORs associated with the multiple-coincidence data, wherein a respective part of the multiple-coincidence data allocated to each of the LORs is respective multiple allocation data on the LOR;

correcting, for each of the LORs, respective double-coincidence data in the coincidence data corresponding to the LOR based on the respective multiple allocation data on the LOR to obtain respective data of the LOR; and

reconstructing an image according to the respective data of each of the LORs.

2. The method of claim 1 , wherein obtaining the respective line integral value along the LOR comprises:

determining a Time-of-Flight (TOF) kernel function according to the time difference between the two single events corresponding to the LOR; and

integrating the TOF kernel function and pixel values of a second image in a corresponding range along the LOR to obtain the line integral value of the LOR associated with the multiple-coincidence data, wherein the second image is reconstructed according to the double-coincidence data in the coincidence data.

3. The method of claim 2 , wherein the corresponding range is determined based on the TOF kernel function and has a width larger than a width of an event occurrence region on the LOR that is determined based on the time difference between the two single events.

4. The method of claim 2 , wherein determining the TOF kernel function according to the time difference between the two single events comprises:

obtaining a function center of the TOF kernel function according to the time difference between the two single events; and

determining the TOF kernel function according to the function center.

5. The method of claim 2 , wherein determining the TOF kernel function according to the time difference between the two single events comprises:

obtaining a midpoint of a time difference region corresponding to the time difference between the two single events;

obtaining a function center of the TOF function according to the midpoint; and

determining the TOF kernel function according to the function center.

6. The method of claim 1 , wherein obtaining the respective line integral value along the LOR comprises:

dividing the LOR into a plurality of event occurrence regions based on time difference regions;

performing TOF orthographic projection for a second image to obtain a respective line integral value corresponding to each of the plurality of event occurrence regions on the LOR, wherein the second image is reconstructed according to the double-coincidence data in the coincidence data;

generating correspondences between the time difference regions and the respective line integral values corresponding to the plurality of event occurrence regions on the LOR; and

obtaining the line integral value along the LOR according to the time difference between the two single events corresponding to the LOR and the correspondences.

7. The method of claim 1 , wherein allocating the multiple-coincidence data to the LORs according to the obtained respective line integral values of the LORs associated with the multiple-coincidence data comprises:

allocating a respective proportion of a count value of the multiple-coincidence data to each of the LORs according to the respective line integral values of the LORs associated with the multiple-coincidence data.

8. The method of claim 1 , wherein correcting the respective double-coincidence data in the coincidence data corresponding to the LOR based on the respective multiple allocation data on the LOR comprises:

adding a count value of the multiple allocation data on the LOR and a count value of the double-coincidence data corresponding to the LOR together to obtain a corrected count value; and

taking the corrected count value as a count value of the data of the LOR.

9. The method of claim 1 , wherein reconstructing the image according to the respective data of each of the LORs comprises:

generating a scatter correction factor and an attenuation correction factor according to the respective data of each of the LORs; and

reconstructing the image according to the scatter correction factor, the attenuation correction factor and the respective data of each of the LORs.

10. A PET device, comprising:

a processor, and

a non-transitory machine-readable storage medium storing machine executable instructions which are executable by the processor to:

determine a plurality of Lines of Response (LORs) associated with multiple-coincidence data in coincidence data detected by a PET device through scanning, wherein the coincidence data comprises double-coincidence data indicating that two single events satisfy a coincidence condition and the multiple-coincidence data indicating that three or more single events satisfy the coincidence condition, and each of the LORs is a line between crystals corresponding to every two of the three or more single events corresponding to the multiple-coincidence data;

obtain, for each of the LORs, a respective line integral value along the LOR according to a time difference between two single events corresponding to the LOR;

allocate the multiple-coincidence data to the LORs according to the obtained respective line integral values of the LORs associated with the multiple-coincidence data, wherein a respective part of the multiple-coincidence data allocated to each of the LORs is respective multiple allocation data on the LOR;

correct, for each of the LORs, respective double-coincidence data in the coincidence data corresponding to the LOR based on the respective multiple allocation data on the LOR to obtain respective data of the LOR; and

reconstruct an image according to the respective data of each of the LORs.

11. The device of claim 10 , wherein the processor is caused by the machine-executable instructions to:

determine a Time-of-Flight (TOF) kernel function according to the time difference between the two single events corresponding to the LOR; and

integrate the TOF kernel function and pixel values of a second image in a corresponding range along the LOR to obtain the line integral value of the LOR associated with the multiple-coincidence data, wherein the second image is reconstructed according to the double-coincidence data in the coincidence data.

12. The device of claim 11 , wherein the processor is caused by the machine-executable instructions to:

obtain a function center of the TOF kernel function according to the time difference between the two single events; and

determine the TOF kernel function according to the function center.

13. The device of claim 11 , wherein the processor is caused by the machine-executable instructions to:

obtain a midpoint of a time difference region corresponding to the time difference between the two single events;

obtain a function center of the TOF function according to the midpoint; and

determine the TOF kernel function according to the function center.

14. The device of claim 10 , wherein the processor is caused by the machine-executable instructions to:

divide the LOR into a plurality of event occurrence regions based on time difference regions;

perform TOF orthographic projection for a second image to obtain a respective line integral value corresponding to each of the plurality of the event occurrence regions on the LOR, wherein the second image is reconstructed according to the double-coincidence data in the coincidence data;

generate correspondences between the time difference regions and the respective line integral values corresponding to the plurality of event occurrence regions on the LOR;

obtain the line integral value along the LOR according to the time difference between the two single events corresponding to the LOR and the correspondences.

15. The device of claim 10 , wherein the processor is caused by the machine-executable instructions to:

allocate a respective proportion of count value of the multiple-coincidence data to each of the LORs according to the respective line integral values of the LORs associated with the multiple-coincidence data.

16. The device of claim 10 , wherein the processor is caused by the machine-executable instructions to:

add a count value of the multiple allocation data on the LOR and a count value of the double-coincidence data corresponding to the LOR together to obtain a corrected count value; and

take the corrected count value as a count value of the data of the LOR.

17. The device of claim 10 , wherein the processor is caused by the machine-executable instructions to:

generate a scatter correction factor and an attenuation correction factor according to the respective data of each of the LORs; and

reconstruct the image according to the scatter correction factor, the attenuation correction factor and the respective data of each of the LORs.

18. A method of reconstructing a Positron Emission Computed Tomography (PET) image, comprising:

reconstructing a first PET image according to double coincidence data in coincidence data detected by a PET device through scanning;

determining a plurality of lines of response (LOR) associated with multiple-coincidence data in the coincidence data, each of the LORs being a line between crystals corresponding to every two of three or more single events corresponding to the multiple-coincidence data;

determining a respective weight for each of the LORs based on a time difference between two single events corresponding to the LOR and an overlapping region between the first PET image and the LOR;

allocating the multiple-coincidence data to the LORs based on the determined respective weights to obtain respective multiple allocation data on the LORs;

correcting, for each of the LORs, respective double-coincidence data in the coincidence data corresponding to the LOR based on the respective multiple allocation data on the LOR to obtain respective data of the LOR; and

reconstructing a second PET image according to the respective data of the LORs.

19. The method of claim 18 , wherein determining a respective weight for each of the LORs comprises:

determining a Time-of-Flight (TOF) kernel function according to the time difference between the two single events corresponding to the LOR; and

integrating the TOF kernel function and pixel values of the first PET image in an integral range along the LOR to obtain a line integral value of the LOR associated with the multiple-coincidence data, wherein the line integral value is determined to be the respective weight for the LOR.

20. The method of claim 18 , wherein the integral range is determined based on the TOF kernel function and has a width larger than a width of an event occurrence region on the LOR that is determined based on the time difference between the two single events.