Granted Patent B2

US 12,211,176 · App. 17/645,037 · Granted Jan 28, 2025

Inventors: Liang Xu (Shanghai, CN); Juan Feng (Shanghai, CN); Jun Xiang (Shanghai, CN); Na Zhang (Shanghai, CN); Yang Hu (Shanghai, CN); Yan'ge Ma (Shanghai, CN); Kai Cui (Shanghai, CN); Le Yang (Shanghai, CN)

Assignee: SHANGHAI UNITED IMAGING HEALTHCARE CO., LTD.

G06T5/10G06T5/70G06V10/44G06T2207/10072G06V10/473G06V2201/03

Abstract

Claims (83)

1. A system for image processing, comprising:

at least one storage device including a set of instructions; and

at least one processor in communication with the at least one storage device, wherein when executing the set of instructions, the at least one processor is directed to perform operations including:

obtaining at least one image of an object;

for each of the at least one image, determining a recognition result of the image, wherein the recognition result includes at least one of an image type of the image, a type of a lesion in the image, a region of the lesion in the image, or an image feature of the image; and

processing the at least one image of the object based on at least one recognition result corresponding to the at least one image, wherein each of the at least one image is a three-dimensional (3D) tomographic image, the processing the at least one image of the object includes:

determining at least one target gradient image based on at least one 3D tomographic image;

generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image.

2. The system of claim 1 , wherein

the generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image includes:

determining a target image of the object by preprocessing each of the at least one 3D tomographic image;

determining, in each of the at least one target gradient image, a plurality of first positions satisfying a predetermined gradient condition along a predetermined direction;

determining, in the target image, a plurality of second positions corresponding to the plurality of first positions; and

generating an initial two-dimensional (2D) image of the object based on gray values of the plurality of second positions in the target image, and

generating the target 2D image of the object based on at least one initial 2D image of the object corresponding to the at least one 3D tomographic image.

3. The system of claim 2 , wherein the determining the target image of the object by preprocessing each of the at least one 3D tomographic image includes:

generating at least one high-frequency image by enhancing high-frequency information in the 3D tomographic image; and

generating the target image of the object based on the at least one high-frequency image and a low-frequency image corresponding to low-frequency information in the 3D tomographic image.

4. The system of claim 3 , wherein the generating the at least one high-frequency image by enhancing the high-frequency information in the 3D tomographic image includes:

generating a plurality of first frequency band images by decomposing the 3D tomographic image based on a first predetermined frequency band, wherein the plurality of first frequency band images include a low-frequency image and at least one initial high-frequency image; and

generating the at least one high-frequency image by enhancing the at least one initial high-frequency image in the plurality of first frequency band images.

5. The system of claim 2 , wherein the determining the target gradient image includes:

generating an initial gradient image by performing a gradient transformation on the target image; and

determining the target gradient image by performing a convolution processing on the initial gradient image.

6. The system of claim 2 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating the target 2D image of the object by projecting the at least one initial 2D image along the predetermined direction.

7. The system of claim 2 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating a first projected image of the object by projecting the at least one initial 2D image along the predetermined direction based on a maximum density projection algorithm;

generating a second projected image of the object by projecting the at least one initial 2D image along the predetermined direction based on an average projection algorithm; and

generating the target 2D image of the object by weighting the first projected image and the second projected image based on a first predetermined weight of the first projected image and a second predetermined weight of the second projected image.

8. The system of claim 2 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating a preliminary target 2D image of the object by combining the at least one initial 2D image; and

generating the target 2D image of the object by weighting the preliminary target 2D image and a 2D projected image based on a third predetermined weight of the preliminary target 2D image and a fourth predetermined weight of the 2D projected image.

9. The system of claim 2 , wherein the operations further include:

generating a plurality of second frequency band images by decomposing the target 2D image based on a second predetermined frequency band, wherein the plurality of second frequency band images include a low-frequency 2D image and a high-frequency 2D image;

generating a high frequency enhanced image by enhancing the high-frequency 2D image in the second frequency band image; and

correcting the target 2D image based on the high frequency enhanced image.

10. The system of claim 1 , wherein the processing the at least one image of the object based on the at least one recognition result includes:

for each of the at least one image,

determining a processing operation based on the type of the lesion in the image or the image feature of the image, wherein the processing operation includes at least one of a noise reduction, an enhancement, an organization balance, a thickness balance, or a look-up table (LUT) transformation; and

processing the image based on the processing operation.

11. The system of claim 1 , wherein the operations further include:

providing prompt information associated with the recognition result of the image in the processed image, the prompt information including at least one of highlighting the recognition result, zooming in the recognition result, circling the recognition result, or prompting the recognition result using a prompt indicator.

12. The system of claim 1 , wherein the generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image includes:

determining a target image of the object by preprocessing each of the at least one 3D tomographic image;

determining, in each of the at least one target gradient image, a plurality of first positions satisfying a predetermined gradient condition along a predetermined direction;

determining, in the target image, a plurality of second positions corresponding to the plurality of first positions; and

generating the target 2D image of the object based on the plurality of second positions in each of the at least one target image.

13. A method implemented on a computing device including at least one processor, at least one storage medium, and a communication platform connected to a network, the method comprising:

obtaining at least one image of an object;

for each of the at least one image, determining a recognition result of the image, wherein the recognition result includes at least one of an image type of the image, a type of a lesion in the image, a region of the lesion in the image, or an image feature of the image; and

processing the at least one image of the object based on at least one recognition result corresponding to the at least one image, wherein each of the at least one image is a three-dimensional (3D) tomographic image, the processing the at least one image of the object includes:

determining at least one target gradient image based on at least one 3D tomographic image;

generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image.

14. The method of claim 13 , wherein

the generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image includes:

determining a target image of the object by preprocessing each of the at least one 3D tomographic image;

determining, in each of the at least one target gradient image, a plurality of first positions satisfying a predetermined gradient condition along a predetermined direction;

determining, in the target image, a plurality of second positions corresponding to the plurality of first positions; and

generating an initial two-dimensional (2D) image of the object based on gray values of the plurality of second positions in the target image, and

generating the target 2D image of the object based on at least one initial 2D image of the object corresponding to the at least one 3D tomographic image.

15. The method of claim 14 , wherein the determining the target gradient image includes:

generating an initial gradient image by performing a gradient transformation on the target image; and

determining the target gradient image by performing a convolution processing on the initial gradient image.

16. The method of claim 14 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating the target 2D image of the object by projecting the at least one initial 2D image along the predetermined direction.

17. The method of claim 14 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating a first projected image of the object by projecting the at least one initial 2D image along the predetermined direction based on a maximum density projection algorithm;

generating a second projected image of the object by projecting the at least one initial 2D image along the predetermined direction based on an average projection algorithm; and

generating the target 2D image of the object by weighting the first projected image and the second projected image based on a first predetermined weight of the first projected image and a second predetermined weight of the second projected image.

18. The method of claim 14 , wherein the generating the target 2D image of the object based on the at least one initial 2D image of the object includes:

generating a preliminary target 2D image of the object by combining the at least one initial 2D image; and

generating the target 2D image of the object by weighting the preliminary target 2D image and a 2D projected image based on a third predetermined weight of the preliminary target 2D image and a fourth predetermined weight of the 2D projected image.

19. The method of claim 14 , wherein the method further include:

generating a plurality of second frequency band images by decomposing the target 2D image based on a second predetermined frequency band, wherein the plurality of second frequency band images include a low-frequency 2D image and a high-frequency 2D image;

generating a high frequency enhanced image by enhancing the high-frequency 2D image in the second frequency band image; and

correcting the target 2D image based on the high frequency enhanced image.

20. A non-transitory computer readable medium, comprising executable instructions that, when executed by at least one processor, direct the at least one processor to perform a method, the method comprising:

obtaining at least one image of an object;

for each of the at least one image, determining a recognition result of the image, wherein the recognition result includes at least one of an image type of the image, a type of a lesion in the image, a region of the lesion in the image, or an image feature of the image; and

processing the at least one image of the object based on at least one recognition result corresponding to the at least one image, wherein each of the at least one image is a three-dimensional (3D) tomographic image, the processing the at least one image of the object includes:

determining at least one target gradient image based on at least one 3D tomographic image;

generating a target two-dimensional (2D) image of the object based on the at least one 3D tomographic image and the at least one target gradient image.

Assignments (1)

Priority Claims (3)

Continuity (2)

References Cited (37)