Reconstructing images

1. A method of reconstructing an image, comprising:

obtaining scanning data generated by scanning a subject;

processing the scanning data to generate an initially-updated image;

generating image boundary prior information by performing at least two sparse transforms on the initially-updated image; and

performing a weighted reconstruction on the image boundary prior information and the initially-updated image to generate a reconstructed image.

2. The method according to claim 1 , wherein the at least two sparse transforms comprise a gradient transform and a wavelet transform.

3. The method according to claim 1 , wherein generating the image boundary prior information comprises:

performing a gradient transform on the initially-updated image to generate a PET gradient image;

performing a wavelet transform on the PET gradient image to generate first sub-images in a wavelet domain;

denoising the first sub-images in the wavelet domain to generate denoised first sub-images in the wavelet domain; and

performing a wavelet inverse transform on the denoised first sub-images in the wavelet domain to generate an updated gradient image,

wherein the updated gradient image comprises the image boundary prior information.

4. The method according to claim 3 , wherein the weighted reconstruction performed on the image boundary prior information and the initially-updated image is iterative and is based on an iterative update formula, wherein the image boundary prior information functions as a weighting factor to iteratively update the initially-updated image.

5. The method according to claim 1 , wherein generating the image boundary prior information comprises:

performing a wavelet transform on the initially-updated image to generate initial sub-images in a wavelet domain;

denoising the initial sub-images in the wavelet domain to generate denoised initial sub-images in the wavelet domain;

performing a wavelet inverse transform on the denoised initial sub-images in the wavelet domain to generate an initial boundary image;

performing a gradient transform on the initial boundary image to generate an updated gradient image; and

wherein the updated gradient image comprises the image boundary prior information.

6. The method according to claim 5 , wherein the weighted reconstruction performed on the image boundary prior information and the initially-updated image is iterative and is based on an iterative update formula, wherein the image boundary prior information functions as a weighting factor to iteratively update the initially-updated image.

7. The method according to claim 1 , wherein generating the image boundary prior information comprises:

performing a gradient transform on the initially-updated image to generate a PET gradient image;

performing a wavelet transform on the PET gradient image to generate first sub-images in a wavelet domain;

denoising the first sub-images in the wavelet domain to generate denoised first sub-images in the wavelet domain; and

wherein the denoised first sub-images in the wavelet domain comprises the image boundary prior information.

8. The method according to claim 7 , wherein the weighted reconstruction performed on the image boundary prior information in the initially-updated image comprises:

performing a wavelet transform on the initially-updated image to generate second sub-images in the wavelet domain;

performing a weighted stacking operation on each of the denoised first sub-images with one of the second sub-images at corresponding frequency bands in the wavelet domain to generate weighted PET wavelet sub-images; and

performing a wavelet inverse transform on the weighted PET wavelet sub-images to generate the reconstructed image.

9. The method according to claim 7 , wherein the weighted reconstruction performed on the image boundary prior information in the initially-updated image comprises:

denoising the initially-updated image;

performing a wavelet transform on the denoised initially-updated image to generate third sub-images in the wavelet domain;

performing a weighted stacking operation on each of the denoised first sub-images with one of the third sub-images at corresponding frequency bands in the wavelet domain to generate weighted PET wavelet sub-images; and

performing a wavelet inverse transform on the weighted PET wavelet sub-images to generate the reconstructed image.

10. A device for reconstructing an image, comprising:

a processor configured to execute machine readable instructions encoding control logic for reconstructing the image, the instructions being read from a machine readable storage medium, wherein executing the machine readable instructions cause the processor to:

obtain scanning data generated by scanning a subject;

process the scanning data to generate an initially-updated image;

generate image boundary prior information by performing at least two sparse transforms on the initially-updated image; and

perform a weighted reconstruction on the image boundary prior information and the initially-updated image to generate a reconstructed image.

11. The device of claim 10 , wherein the at least two sparse transforms comprise a gradient transform and a wavelet transform.

12. The device of claim 10 , wherein when generating the image boundary prior information, the machine executable instructions cause the processor to:

perform a gradient transform on the initially-updated image to generate a PET gradient image;

perform a wavelet transform on the PET gradient image to generate first sub-images in a wavelet domain;

denoise the first sub-images in the wavelet domain to generate denoised first sub-images in the wavelet domain; and

perform a wavelet inverse transform on the denoised first sub-images in the wavelet domain to generate an updated gradient image,

wherein the updated gradient image comprises the image boundary prior information.

13. The device of claim 12 , wherein the weighted reconstruction performed on the image boundary prior information and the initially-updated image is iterative and is based on an iterative update formula, wherein the image boundary prior information functions as a weighting factor to iteratively update the initially-updated image.

14. The device of claim 10 , wherein when performing the weighted reconstruction on the image boundary prior information and the initially-updated image, the machine executable instructions cause the processor to:

perform a wavelet transform on the initially-updated image to generate initial sub-images in a wavelet domain;

denoise the initial sub-images in the wavelet domain to generate denoised initial sub-images in the wavelet domain;

perform a wavelet inverse transform on the denoised initial sub-images in the wavelet domain to generate an initial boundary image; and

perform a gradient transform on the initial boundary image to generate an updated gradient image,

wherein the updated gradient image comprises the image boundary prior information.

15. The device of claim 14 , wherein the weighted reconstruction performed on the image boundary prior information and the initially-updated image is iterative and is based on an iterative update formula, wherein the image boundary prior information functions as a weighting factor to iteratively update the initially-updated image.

16. The device of claim 10 , wherein when generating the image boundary prior information, the machine executable instructions cause the processor to:

perform a gradient transform on the initially-updated image to generate a PET gradient image;

perform a wavelet transform on the PET gradient image to generate first sub-images in a wavelet domain; and

denoising the first sub-images in the wavelet domain to generate denoised first sub-images in the wavelet domain,

wherein the denoised first sub-images comprises the wavelet domain as the image boundary prior information.

17. The device of claim 16 , wherein when performing the weighted reconstruction on the image boundary prior information and the initially-updated image, the machine executable instructions cause the processor to:

perform a wavelet transform on the initially-updated image to generate second sub-images in the wavelet domain;

perform a weighted stacking operation on each of the denoised first sub-images with one of the second sub-images at corresponding frequency bands in the wavelet domain to generate weighted PET wavelet sub-images; and

perform a wavelet inverse transform on the weighted PET wavelet sub-images to generate the reconstructed image.

18. The device of claim 16 , wherein when performing the weighted reconstruction on the image boundary prior information and the initially-updated image, the machine executable instructions cause the processor to:

denoise the initially-updated image;

perform a wavelet transform on the denoised initially-updated image to generate third sub-images in the wavelet domain;

perform a weighted stacking operation on each of the denoised first sub-images with one of the third sub-images at corresponding frequency bands in the wavelet domain to generate weighted PET wavelet sub-images; and

perform a wavelet inverse transform on the weighted PET wavelet sub-images to generate the reconstructed image.

19. A machine-readable storage medium coupled to a processor and having machine-executable instructions stored thereon that, when executed by the processor, cause the processor to perform operations comprising:

obtaining scanning data generated by scanning a subject;

processing the scanning data to generate an initially-updated image;

generating image boundary prior information by performing at least two sparse transforms on the initially-updated image; and

performing a weighted reconstruction on the image boundary prior information and the initially-updated image to generate a reconstructed image.