Granted Patent B2

US 12,347,107 · App. 17/974,453 · Granted Jul 1, 2025

Inventors: Liang Wang (Shenzhen, CN); Jianhua Yao (Shenzhen, CN)

Assignee: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED

G06T7/0014G06T7/11G06T7/149H04N1/6005G06T2207/10024G06T2207/20224G06T2207/30068G06T2207/30096

Abstract

Claims (91)

1. A medical image processing method performed by a computer device, comprising:

obtaining a medical image set, the medical image set including a reference medical image, a target medical image to be identified, and annotated area size information of a non-lesion region in the target medical image, the target medical image including a lesion region and the non-lesion region, and the reference medical image including a lesion region;

identifying one or more differences between the reference medical image and the target medical image to obtain a candidate non-lesion region in the target medical image;

determining area size information of the candidate non-lesion region as candidate area size information; and

adjusting the candidate non-lesion region according to the annotated area size information when the candidate area size information does not match the annotated area size information to obtain a target non-lesion region in the target medical image.

2. The method according to claim 1 , wherein identifying one or more differences between the reference medical image and the target medical image includes:

converting the reference medical image to an image in a target color space to obtain a converted reference medical image,

converting the target medical image to an image in the target color space to obtain a converted target medical image;

obtaining brightness information of pixel points in the converted reference medical image and brightness information of pixel points in the converted target medical image; and

identifying the one or more differences between the reference medical image and the target medical image according to the brightness information of the pixel points in the converted reference medical image and the brightness information of the pixel points in the converted target medical image, to obtain a candidate non-lesion region in the target medical image.

3. The method according to claim 2 , wherein identifying the one or more differences between the reference medical image and the target medical image according to the brightness information of the pixel points in the converted reference medical image and the brightness information of the pixel points in the converted target medical image includes:

obtaining a difference between the brightness information of the pixel points in the converted reference medical image and the brightness information of the corresponding pixel points in the converted target medical image;

identifying pixel points of which the difference corresponding to brightness information is greater than a difference threshold from the converted target medical image as first target pixel points; and

determining a region where the first target pixel points are located in the target medical image as the candidate non-lesion region in the target medical image.

4. The method according to claim 1 , wherein the candidate area size information comprises a candidate area proportion of the candidate non-lesion region in the target medical image; the annotated area size information comprises a marked area proportion of the non-lesion region in the target medical image;

adjusting the candidate non-lesion region according to the annotated area size information when the candidate region size information does not match the annotated area size information, so as to obtain a target non-lesion region in the target medical image comprises:

obtaining a difference between the candidate area proportion and the marked area proportion;

determining that the candidate area size information is not matched with the annotated area size information when the difference is greater than a proportion threshold; and

adjusting the candidate non-lesion region according to the marked area proportion to obtain the target non-lesion region.

5. The method according to claim 4 , wherein adjusting the candidate non-lesion region according to the annotated area proportion to obtain the target non-lesion region comprises:

expanding the candidate non-lesion region according to the annotated area proportion when the candidate area proportion is less than the marked area proportion, so as to obtain the target non-lesion region; and

reducing the candidate non-lesion region according to the annotated area proportion when the candidate area proportion is greater than the annotated area proportion, so as to obtain the target non-lesion region.

6. The method according to claim 5 , wherein expanding the candidate non-lesion region according to the annotated area proportion when the candidate area proportion is less than the annotated area proportion, so as to obtain the target non-lesion region comprises:

obtaining expanding parameters when the candidate area proportion is less than the annotated area proportion, the expanding parameters comprising an expanding shape and an expanding size;

iteratively expanding the candidate non-lesion region according to the expanding parameters to obtain an expanded candidate non-lesion region;

obtaining a region proportion of the expanded candidate non-lesion region in the target medical image as an expanded region proportion; and

determining the expanded candidate non-lesion region of which the area proportion difference between the expanded region proportion and the annotated area proportion is less than the proportion threshold as the target non-lesion region.

7. The method according to claim 5 , wherein expanding the candidate non-lesion region according to the annotated area proportion when the candidate area proportion is less than the annotated area proportion, so as to obtain the target non-lesion region comprises:

obtaining pixel values of pixel points adjacent to the candidate non-lesion region in the target medical image and pixel values of pixel points in the candidate non-lesion region when the candidate area proportion is less than the annotated area proportion;

clustering the pixel points in the candidate non-lesion region and the pixel points adjacent to the candidate non-lesion region according to the pixel values of the pixel points in the candidate non-lesion region and the pixel values of the pixel points adjacent to the candidate non-lesion region, so as to obtain a clustered area;

obtaining an area proportion of the clustered region in the target medical image as a clustered area proportion; and

determining the clustered region as the target non-lesion region when a difference between the clustered area proportion and the annotated area proportion is less than the proportion threshold.

8. The method according to claim 7 , wherein clustering the pixel points in the candidate non-lesion region and the pixel points adjacent to the candidate non-lesion region according to the pixel values of the pixel points in the candidate non-lesion region and the pixel values of the pixel points adjacent to the candidate non-lesion region, so as to obtain a clustered area comprises:

obtaining a pixel difference between the pixel values of the pixel points adjacent to the candidate non-lesion region and the pixel values of the pixel points in the candidate non-lesion region;

determining the pixel points of which the corresponding pixel difference is less than a pixel difference threshold from the pixel points adjacent to the candidate non-lesion region as second target pixel points; and

merging an area corresponding to pixel locations of the second target pixel points with the candidate non-lesion region to obtain the clustered area.

9. The method according to claim 5 , wherein reducing the candidate non-lesion region according to the annotated area proportion when the candidate region proportion is greater than the annotated area proportion, so as to obtain the target non-lesion region comprises:

obtaining reduction processing parameters when the candidate region proportion is greater than the annotated area proportion, the reduction processing parameters comprising a shape of reduction processing and a size of reduction processing;

iteratively reducing the candidate non-lesion region according to the reduction processing parameters to obtain a reduced candidate non-lesion region;

obtaining a region proportion of the reduced candidate non-lesion region in the target medical image as a reduced area proportion; and

determining the reduced candidate non-lesion region of which the region proportion difference between the reduced area proportion and the annotated area proportion is less than the proportion threshold as the target non-lesion region.

10. The method according to claim 1 , wherein the candidate area size information comprises a candidate area size of the candidate non-lesion region; the annotated area information comprises an annotated area size of the non-lesion region in the target medical image;

adjusting the candidate non-lesion region according to the annotated area size information when the candidate region size information is not matched with the annotated area size information to obtain a target non-lesion region in the target medical image includes:

obtaining an area size difference between the candidate area size and the annotated area size;

determining that the candidate region size information is not matched with the annotated size information when the area size difference is greater than a size threshold; and

adjusting the candidate non-lesion region according to the annotated area size to obtain the target non-lesion region.

11. The method according to claim 4 , wherein determining the area size information of the candidate non-lesion region as the candidate area size information comprises:

obtaining an area size of the candidate non-lesion region and an image size of the target medical image, and determining a ratio of the area size of the candidate non-lesion region to the image size of the target medical image as the candidate area proportion; or,

obtaining the number of the pixel points in the candidate non-lesion region and the number of the pixel points in the target medical image, and determining a ratio of the number of the pixel points in the candidate non-lesion region to the number of the pixel points in the target medical image as the candidate region proportion; and

determining the candidate area proportion as the candidate area size information.

12. The method according to claim 1 , wherein the method further comprises:

marking the target non-lesion region in the target medical image to obtain a marked target medical image;

predicting the target medical image by an image segmentation model to obtain a predicted non-lesion region in the target medical image, and marking the predicted non-lesion region in the target medical image to obtain a predicted target medical image; and

adjusting the image segmentation model according to the marked target medical image and the predicted target medical image to obtain a target medical image segmentation model.

13. The method according to claim 12 , wherein adjusting the image segmentation model includes:

determining a predicted loss value of the image segmentation model according to the marked target medical image and the predicted target medical image; and

adjusting the image segmentation model according to the predicted loss value when the predicted loss value does not meet a convergence condition, so as to obtain a target medical image segmentation model.

14. The method according to claim 12 , wherein marking the target non-lesion region includes:

binarizing the target medical image according to the pixel points in the target non-lesion region; and

determining the binarized target medical image as the marked target medical image.

15. A computer device, comprising:

one or more processors; and

memory storing one or more programs, the one or more programs comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:

obtaining a medical image set, the medical image set including a reference medical image, a target medical image to be identified, and annotated area size information of a non-lesion region in the target medical image, the target medical image including a lesion region and the non-lesion region, and the reference medical image including a lesion region;

identifying one or more differences between the reference medical image and the target medical image to obtain a candidate non-lesion region in the target medical image;

determining area size information of the candidate non-lesion region as candidate area size information; and

adjusting the candidate non-lesion region according to the annotated area size information when the candidate area size information does not match the annotated area size information, so as to obtain a target non-lesion region in the target medical image.

16. The computer device according to claim 15 , wherein identifying one or more differences between the reference medical image and the target medical image includes:

converting the reference medical image to an image in a target color space to obtain a converted reference medical image,

converting the target medical image to an image in the target color space to obtain a converted target medical image;

obtaining brightness information of pixel points in the converted reference medical image and brightness information of pixel points in the converted target medical image; and

identifying the one or more differences between the reference medical image and the target medical image according to the brightness information of the pixel points in the converted reference medical image and the brightness information of the pixel points in the converted target medical image, so as to obtain a candidate non-lesion region in the target medical image.

17. The computer device according to claim 16 , wherein identifying the one or more differences between the reference medical image and the target medical image according to the brightness information of the pixel points in the converted reference medical image and the brightness information of the pixel points in the converted target medical image includes:

obtaining a difference between the brightness information of the pixel points in the converted reference medical image and the brightness information of the corresponding pixel points in the converted target medical image;

identifying pixel points of which the difference corresponding to brightness information is greater than a difference threshold from the converted target medical image as first target pixel points; and

determining a region where the first target pixel points are located in the target medical image as the candidate non-lesion region in the target medical image.

18. The computer device according to claim 15 , wherein the candidate area size information comprises a candidate area proportion of the candidate non-lesion region in the target medical image; the annotated area size information comprises a marked area proportion of the non-lesion region in the target medical image;

adjusting the candidate non-lesion region according to the annotated area size information when the candidate region size information does not match the annotated area size information, so as to obtain a target non-lesion region in the target medical image comprises:

obtaining an area proportion difference between the candidate area proportion and the marked area proportion;

determining that the candidate area size information is not matched with the annotated area size information when the area proportion difference is greater than a proportion threshold; and

adjusting the candidate non-lesion region according to the marked area proportion to obtain the target non-lesion region.

19. The computer device according to claim 15 , wherein the candidate area size information comprises a candidate area size of the candidate non-lesion region; the annotated area information comprises an annotated area size of the non-lesion region in the target medical image;

adjusting the candidate non-lesion region according to the annotated area size information when the candidate region size information is not matched with the annotated area size information to obtain a target non-lesion region in the target medical image includes:

obtaining an area size difference between the candidate area size and the annotated area size;

determining that the candidate region size information is not matched with the annotated size information when the area size difference is greater than a size threshold; and

adjusting the candidate non-lesion region according to the annotated area size to obtain the target non-lesion region.

20. A non-transitory computer-readable storage medium, storing a computer program, the computer program, when executed by one or more processors of a computer device, cause the one or more processors to perform operations comprising:

obtaining a medical image set, the medical image set including a reference medical image, a target medical image to be identified, and annotated area size information of a non-lesion region in the target medical image, the target medical image including a lesion region and the non-lesion region, and the reference medical image including a lesion region;

identifying one or more differences between the reference medical image and the target medical image to obtain a candidate non-lesion region in the target medical image;

determining area size information of the candidate non-lesion region as candidate area size information; and

adjusting the candidate non-lesion region according to the annotated area size information when the candidate area size information does not match the annotated area size information, so as to obtain a target non-lesion region in the target medical image.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (16)