Medical image segmentation method and apparatus, computer device, and storage medium

1. A medical image segmentation method, applicable to a computer device, the method comprising:

obtaining medical image data;

obtaining a target object and weakly supervised annotation information of the target object in the medical image data;

determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information; and

performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object, further comprising:

performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object; and

fusing the at least two intermediate segmentation results, to obtain the target segmentation result for the target object.

2. The method according to claim 1 , wherein the performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object comprises:

performing mapping on the medical image data based on the pseudo segmentation mask and by using at least two neural network models with no less than two structural types, to obtain the at least two intermediate segmentation results for the target object.

3. The method according to claim 1 , wherein in a training process of the at least two preset mapping models, the at least two preset mapping models are trained in a self-paced learning manner.

4. The method according to claim 3 , wherein the training process of the at least two preset mapping models comprises:

training at least two intermediate preset mapping models based on the pseudo segmentation mask in the training process;

performing mapping on the medical image data in the training process by using the at least two intermediate preset mapping models, to obtain at least two predicted training segmentation results;

generating an advanced pseudo segmentation mask according to the predicted training segmentation results in the at least two predicted training segmentation results that meet a confidence condition in combination with a priori knowledge; and

retraining the at least two intermediate preset mapping models based on the advanced pseudo segmentation mask until a preset stop condition is met, to obtain the preset mapping models.

5. The method according to claim 1 , wherein the performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object comprises:

mapping the medical image data by using the preset mapping model obtained based on the pseudo segmentation mask, to obtain the target segmentation result for the target object.

6. The method according to claim 1 , wherein the performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object comprises:

performing mapping on the medical image data based on the pseudo segmentation mask to obtain a predicted segmentation result for the target object; and

correcting the predicted segmentation result according to the a priori knowledge and/or the weakly supervised annotation information, to obtain the target segmentation result of the target object.

7. The method according to claim 1 , wherein the determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information comprises:

determining the pseudo segmentation mask for the target object in the medical image data based on the a priori knowledge and the weakly supervised annotation information.

8. The method according to claim 1 , wherein the weakly supervised annotation information comprises bounding box information, apostrophe annotation information, or point annotation information.

9. The method according to claim 1 , wherein the method is implemented by using a neural network model based on medical image segmentation.

10. The method according to claim 9 , wherein a training process of the neural network model based on medical image segmentation comprises:

obtaining a target training object, and obtaining training samples based on the target training object, the training samples comprising medical image training data and weakly supervised annotation training information;

determining a pseudo segmentation training mask for the target training object in the medical image training data based on the a priori knowledge and the weakly supervised annotation information; and

obtaining the neural network models through training based on the pseudo segmentation training mask and the medical image training data.

11. The method according to claim 10 , wherein the obtaining the neural network models through training based on the pseudo segmentation training mask and the medical image training data comprises:

obtaining a preset mapping model of the neural network model through training based on the pseudo segmentation training mask;

inputting the medical image training data to the preset mapping models, to obtain predicted training segmentation results;

generating an advanced pseudo segmentation training mask according to the predicted training segmentation results that meet a confidence condition in combination with a priori knowledge; and

retraining the preset mapping model according to the advanced pseudo segmentation training mask, until a preset stop condition is met.

12. A computer device, comprising a memory and a processor, the memory storing a plurality of computer programs that, when executed by the processor, cause the computer device to perform a plurality of operations including:

obtaining medical image data;

obtaining a target object and weakly supervised annotation information of the target object in the medical image data;

determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information; and

performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object, further comprising:

performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object; and

fusing the at least two intermediate segmentation results, to obtain the target segmentation result for the target object.

13. The computer device according to claim 12 , wherein the performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object comprises:

performing mapping on the medical image data based on the pseudo segmentation mask and by using at least two neural network models with no less than two structural types, to obtain the at least two intermediate segmentation results for the target object.

14. The computer device according to claim 12 , wherein the performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object comprises:

mapping the medical image data by using the preset mapping model obtained based on the pseudo segmentation mask, to obtain the target segmentation result for the target object.

15. The computer device according to claim 12 , wherein the performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object comprises:

performing mapping on the medical image data based on the pseudo segmentation mask to obtain a predicted segmentation result for the target object; and

correcting the predicted segmentation result according to the a priori knowledge and/or the weakly supervised annotation information, to obtain the target segmentation result of the target object.

16. The computer device according to claim 12 , wherein the determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information comprises:

determining the pseudo segmentation mask for the target object in the medical image data based on the a priori knowledge and the weakly supervised annotation information.

17. The computer device according to claim 12 , wherein the weakly supervised annotation information comprises bounding box information, apostrophe annotation information, or point annotation information.

18. A non-transitory computer-readable storage medium, storing a plurality of computer programs that, when executed by a processor of a computer device, cause the computer device to perform a plurality of operations including:

obtaining medical image data;

obtaining a target object and weakly supervised annotation information of the target object in the medical image data;

determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information; and

performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object;

performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object, further comprising:

performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object; and

fusing the at least two intermediate segmentation results, to obtain the target segmentation result for the target object.

19. The non-transitory computer-readable storage medium according to claim 18 , wherein the performing mapping on the medical image data based on the pseudo segmentation mask by using at least two preset mapping models, to obtain at least two intermediate segmentation results for the target object comprises:

performing mapping on the medical image data based on the pseudo segmentation mask and by using at least two neural network models with no less than two structural types, to obtain the at least two intermediate segmentation results for the target object.

20. The non-transitory computer-readable storage medium according to claim 18 , wherein the determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information comprises:

determining the pseudo segmentation mask for the target object in the medical image data based on the a priori knowledge and the weakly supervised annotation information.