Self-training learning method and system for medical image segmentation and electronic equipment

Abstract

The invention provides a self-training learning method and system for medical image segmentation and electronic equipment, and relates to the technical field of image processing. The method comprises the following steps: firstly, acquiring original training including benign lesion segmentation training data and malignant lesion segmentation training data, and a test data set; then, equally dividing the training data set, and combining the training data set with the test data set for iterative training to form a plurality of groups of data sets; then, executing a training process by utilizing each group of data set: taking benign data as a main part in the initial stage, gradually increasing the proportion of malignant focus segmentation training data, and training until convergence to obtain a first segmentation model; and predicting a test set by using the model to obtain a focus prediction image. And by comparing the prediction image with the annotation image, generating a false label image of the leak detection area and the false detection area. And adding the pseudo label images into a training set, and training the first segmentation model again until convergence. According to the technical scheme, the segmentation precision and generalization ability of the medical image segmentation model can be improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a self-training learning method, system and electronic equipment for medical image segmentation. Background Art

[0002] At present, the segmentation of lesions such as breast tumors mainly relies on deep learning models, such as convolutional neural networks (CNNs). During the training process, a dataset containing benign and malignant breast tumor images is usually used for training, in the hope that the model can learn the characteristics of distinguishing tumors of different natures.

[0003] However, in actual applications, there are often more samples of benign breast tumors than samples of malignant breast tumors, which may cause the model to over-focus on the characteristics of benign tumors during training, while the ability to identify malignant tumors is relatively weak. In addition, traditional iterative training methods often simply repeat the training process without targeted optimization of the model's weaknesses in each iteration.

[0004] These problems not only affect the segmentation accuracy of medical image segmentation models, but also limit their promotion and effectiveness in clinical applications. Therefore, it is particularly important to develop a training method that can improve the segmentation accuracy and generalization ability of the model. Summary of the invention

[0005] The purpose of this application is to provide a self-training learning method, system and electronic equipment for medical image segmentation, which can improve the segmentation accuracy and generalization ability of the medical image segmentation model.

[0006] This application is implemented as follows:

[0007] In the first aspect, the present application provides a self-training learning method for medical image segmentation, comprising the following steps: obtaining an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data. The original training data set is divided into multiple data sets, and iterative training is performed in combination with the original test data set to form multiple groups of training data sets and test data sets. The training process is performed using each group of training data sets and test data sets formed by iterative training, and the training process includes: gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model; using the first segmentation model to predict the test data set to obtain the corresponding lesion prediction image; by calculating the intersection difference between the lesion prediction image and the annotated image, a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area are obtained respectively; the obtained pseudo-label image of the missed detection area and the pseudo-label image of the false detection area are added to the training set, and the first segmentation model is trained until convergence.

[0008] In some implementations, the dividing the original training data set into multiple data sets, and performing iterative training in combination with the original test data set to form multiple groups of training data sets and test data sets includes: dividing the original training data set into a first training data set and a second training data set according to the number of people; using the original training data set and the original test data set as the first group of training data sets and test data sets; fusing the original test data set and the first training data set to obtain a new test data set, and using the second training data set as the corresponding test data set to obtain a second group of training data sets and test data sets; fusing the original test data set and the second training data set to obtain a new test data set, and using the first training data set as the corresponding test data set to obtain a third group of training data sets and test data sets.

[0009] In some implementations, the method of gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining the first segmentation model includes: using the benign lesion segmentation training data in the training data set of the current group to train the original medical image segmentation model until convergence; using all benign lesion segmentation training data and part of the malignant lesion segmentation training data to train the model, and the malignant lesion segmentation training data is gradually increased in multiple training stages until all are used; using all benign lesion segmentation training data and malignant lesion segmentation training data to train the model until convergence; performing data enhancement processing on all benign lesion segmentation training data and malignant lesion segmentation training data, and using the data after data enhancement processing to train the model until convergence, to obtain the first segmentation model.

[0010] In some implementations, the method calculates the intersection difference between the lesion prediction image and the annotated image to obtain a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area, respectively, including: for each lesion segmentation prediction image in the test set, calculating the intersection with the corresponding annotated image to obtain the common area; subtracting the common area from the annotated image to obtain a pseudo-label image of the missed detection area; and subtracting the common area from the predicted image to obtain a pseudo-label image of the false detection area.

[0011] In some implementations, the pseudo-label images of missed detection areas and falsely detected areas are added to a training set, and the first segmentation model is trained until convergence, including: iteratively updating the pseudo-label images of missed detection areas and the pseudo-label images of falsely detected areas, and in each iteration, retraining the first segmentation model until convergence using the data in the training data set and the updated pseudo-label images of missed detection areas and the pseudo-label images of falsely detected areas; wherein, in each iteration, the performance of the model after each iteration is evaluated using independent test set data, and if the accuracy of the model on the test set is improved compared to the previous iteration, the parameters of the current model are retained as the starting point for the next iteration; and the accuracy of the prediction results of the training data set is calculated using the model after each iteration, and the pseudo-label images of missed detection areas and the pseudo-label images of falsely detected areas are updated / not updated according to the accuracy of the calculated prediction results.

[0012] In some implementations, the accuracy of the prediction results of the training data set is calculated using the model after each iteration, and the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area are updated / not updated according to the accuracy of the calculated prediction results, including: if the accuracy of the fused prediction image is higher than the accuracy of the intermediate prediction image, and higher than the accuracy of the fused prediction image retained in the last update, the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area are updated; otherwise, they are not updated.

[0013] In some implementations, the step of calculating the confidence of the pseudo-label image of the missed detection area and the pseudo-label image of the misdetected area includes: calculating the pseudo-label effectiveness rate, and calculating the pseudo-label confidence by fusing all average accuracies and pseudo-label effectiveness rates.

[0014] In some implementations, the loss calculation of the training process includes: when training is not performed without adding pseudo-label images of missed detection areas and pseudo-label images of false detection areas, calculating the SoftMax cross entropy loss for the intermediate prediction features calculated by the model; when training is performed with pseudo-label images of missed detection areas and pseudo-label images of false detection areas, calculating the SoftMax cross entropy loss for the intermediate prediction features calculated by the model, and calculating the SoftMax cross entropy loss and consistency loss for the fused prediction features, and calculating the SoftMax cross entropy loss or L1 loss for the prediction features of the missed detection areas and the prediction features of the false detection areas.

[0015] In a second aspect, the present application provides a self-training learning system for medical image segmentation, which comprises:

[0016] The data acquisition module is configured to: acquire an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data. The data preprocessing module is configured to: divide the original training data set into multiple data sets, and perform iterative training in combination with the original test data set to form multiple groups of training data sets and test data sets. The model training module is configured to: respectively use each group of training data sets and test data sets formed by iterative training to execute a training process, and the training process includes: gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model; using the first segmentation model to predict the test data set to obtain the corresponding lesion prediction image; by calculating the intersection difference between the lesion prediction image and the annotated image, respectively obtain a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area; adding the obtained pseudo-label image of the missed detection area and the pseudo-label image of the false detection area to the training set, and training the first segmentation model until convergence.

[0017] In a third aspect, the present application provides an electronic device, comprising a memory for storing one or more programs; a processor; when the one or more programs are executed by the processor, a method as described in any one of the first aspects is implemented.

[0018] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0019] The present application proposes a self-training learning method for medical image segmentation, which gradually increases the proportion of malignant lesion segmentation training data so that the model gradually strengthens the recognition of malignant lesion features during the training process, thereby effectively solving the problem of limited recognition ability caused by data imbalance. And it can prevent the propagation of label noise at different stages, while eliminating the need for expensive repository-based label refinement. Furthermore, since the present application gradually increases the proportion of malignant lesion data during the training process and uses pseudo-label images for targeted optimization, the model can gradually learn more refined and accurate lesion features during the training process.

[0020] Among them, the present application also forms multiple sets of training data sets and test data sets through iterative training, and uses different data combinations for training, so that the model can be fully trained under different data distributions, thereby improving the generalization ability of the model.

[0021] In addition, this application generates pseudo-label images of missed and misdetected areas by calculating the intersection difference between the lesion prediction image and the annotated image, and adds these pseudo-label images to the training set for targeted optimization. This strategy not only enables the model to pay attention to its own weaknesses, but also enables the model to be continuously optimized during the training process by continuously iterating and updating the pseudo-label images, thereby improving the segmentation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 A flowchart of an embodiment of a self-training learning method for medical image segmentation according to the present application;

[0024] Figure 2 It is a self-training course learning framework in one embodiment of the present application;

[0025] Figure 3 This is a schematic diagram of the feature fusion process in one embodiment of the present application;

[0026] Figure 4 This is a structural block diagram of an embodiment of a self-training learning system for medical image segmentation of the present application;

[0027] Figure 5 A structural block diagram of an electronic device provided in an embodiment of the present application.

[0028] Icon: 201, processor; 202, memory; 203, communication interface. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0030] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0031] Example 1

[0032] At present, the segmentation of lesions such as breast tumors mainly relies on deep learning models, such as convolutional neural networks (CNNs). In order to improve the adaptability of the model on different data sets, the prior art usually adopts the unsupervised domain adaptation (UDA) method, which focuses on adapting the model trained on the labeled source domain to the unlabeled target domain. Furthermore, the passive domain adaptation (SFDA) method is a more practical setting because it no longer requires access to the source data during adaptation. However, the inventors have found that the latest (SOTA, State of the Art, refers to the model that performs best on a specific task or data set) methods of SFDA mainly focus on self-training based on pseudo-label refinement, which usually has two problems: 1) noisy pseudo-labels are inevitable, which may lead to early training time memory; 2) the refinement process requires maintaining a repository, which will cause a great burden when resources are limited.

[0033] In response to these problems, this application proposes a self-training learning method for medical image segmentation, which can effectively improve the segmentation accuracy and generalization ability of the model by iterative training, gradually increasing the proportion of malignant lesion segmentation training data, and using a weakly supervised pseudo-label confidence optimization method and an optimized pseudo-label update scheme. In addition, this scheme can promote learning from a limited number of pseudo-labels selected based on reliability. This simple and effective step successfully prevents the propagation of label noise at different stages and eliminates the need for expensive repository-based label refinement.

[0034] Specifically, see Figure 1 , a self-training learning method for medical image segmentation includes the following steps: obtaining an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data. The original training data set is evenly divided into multiple data sets, and iterative training is performed in combination with the original test data set to form multiple groups of training data sets and test data sets. The training process is performed using each group of training data sets and test data sets formed by iterative training, and the training process includes: gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model; using the first segmentation model to predict the test data set to obtain a corresponding lesion prediction image; by calculating the intersection difference between the lesion prediction image and the annotated image, a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area are obtained respectively; the obtained pseudo-label image of the missed detection area and the pseudo-label image of the false detection area are added to the training set, and the first segmentation model is trained until convergence.

[0035] In the above embodiment, by adopting multiple strategies such as iterative training, gradually increasing the proportion of malignant data, and using pseudo-label images for targeted optimization, the segmentation accuracy and generalization ability of the model are improved. The process includes: first, obtaining an original training data set containing benign lesion segmentation training data and malignant lesion segmentation training data, and an original test data set for verifying the performance of the model. Then, the original training data set is divided into multiple subsets. The purpose of this is to form multiple groups of different training data sets and test data set combinations in subsequent iterative training to make full use of the data and improve the generalization ability of the model. Then, the iterative training process is performed using the multiple groups of training data sets and test data sets formed above. In each training stage, the proportion of malignant lesion segmentation training data is gradually increased. The purpose of this strategy is to enable the model to gradually adapt to and enhance the recognition ability of malignant lesion features during the training process. Then, through the iteration of multiple training stages, the initial medical image segmentation model is trained until convergence to obtain a first segmentation model. Then, the test data set is predicted using the first segmentation model to obtain the corresponding lesion prediction image. Next, by calculating the intersection difference between the lesion prediction image and the annotated image, the missed detection area and the false detection area of ​​the model can be identified, and pseudo-label images of the missed detection area and the false detection area can be generated respectively. Next, these pseudo-label images can be added to the training set to further optimize the model. By retraining the model until convergence, the segmentation accuracy and generalization ability of the model can be further improved.

[0036] In short, this application gradually increases the proportion of malignant lesion segmentation training data, so that the model gradually strengthens the recognition of malignant lesion features during the training process, thereby effectively solving the problem of limited recognition ability caused by data imbalance. And it can prevent the propagation of label noise at different stages, while eliminating the need for expensive repository-based label refinement. Furthermore, since this application gradually increases the proportion of malignant data during the training process and uses pseudo-label images for targeted optimization, the model can gradually learn more refined and accurate lesion features during the training process.

[0037] Among them, the present application also forms multiple sets of training data sets and test data sets through iterative training, and uses different data combinations for training, so that the model can be fully trained under different data distributions, thereby improving the generalization ability of the model.

[0038] In addition, this application generates pseudo-label images of missed and misdetected areas by calculating the intersection difference between the lesion prediction image and the annotated image, and adds these pseudo-label images to the training set for targeted optimization. This strategy not only enables the model to pay attention to its own weaknesses, but also enables the model to be continuously optimized during the training process by continuously iterating and updating the pseudo-label images, thereby improving the segmentation accuracy.

[0039] Based on the aforementioned scheme, in some implementations of the present application, the original training data set is divided into multiple data sets, and iterative training is performed in combination with the original test data set to form multiple groups of training data sets and test data sets, including: dividing the original training data set into a first training data set and a second training data set according to the number of people; using the original training data set and the original test data set as the first group of training data sets and test data sets; fusing the original test data set and the first training data set to obtain a new test data set, and using the second training data set as the corresponding test data set to obtain a second group of training data sets and test data sets; fusing the original test data set and the second training data set to obtain a new test data set, and using the first training data set as the corresponding test data set to obtain a third group of training data sets and test data sets.

[0040] In the above implementation, the process of dividing the original training data set into multiple data sets and performing iterative training in combination with the original test data set to form multiple training data sets and test data sets is defined and optimized in more detail. The original training data set is first divided into two subsets according to the number of people: the first training data set and the second training data set. This equal division ensures the consistency of the two subsets in terms of data volume and data distribution, providing a basis for subsequent iterative training.

[0041] Then, based on the two equally divided training data sets, combined with the original test data set, iterative training can be performed to form more sets of training data sets and test data sets. Among them, the first set of training data sets and test data sets directly use the original training data set as training data and the original test data set as test data to form the first set of training data sets and test data sets.

[0042] In the construction of the second set of training and test data sets, the original test data set and the first training data set are fused, and the second training data set is used as the corresponding test data set (this is actually a simplification of the expression, to be more precise, the second training data set is used as the training set for the new round of training, and the fused data set is used as the new test set). This fusion and redistribution strategy aims to evaluate and improve the generalization ability of the model by introducing more diverse data.

[0043] In the construction of the third set of training and test data sets, the original test data set and the second training data set are combined to obtain a new test data set. At the same time, the first training data set is used as the training set for this round of iteration. This step further strengthens the model's ability to adapt to diverse data and helps to discover potential weaknesses in the model when processing different data combinations.

[0044] Based on the aforementioned scheme, in some implementations of the present application, the proportion of malignant lesion segmentation training data is gradually increased through multiple training stages, and the initial medical image segmentation model is trained until convergence to obtain a first segmentation model, including: using the benign lesion segmentation training data in the training data set of the current group to train the original medical image segmentation model until convergence; using all benign lesion segmentation training data and part of the malignant lesion segmentation training data to train the model, and the malignant lesion segmentation training data is gradually increased in multiple training stages until all are used; using all benign lesion segmentation training data and malignant lesion segmentation training data to train the model until convergence; performing data enhancement processing on all benign lesion segmentation training data and malignant lesion segmentation training data, and using the data after data enhancement processing to train the model until convergence, to obtain the first segmentation model.

[0045] In the above implementation, at the initial stage of training, only the benign lesion segmentation training data in the current grouped training data set is used to train the original medical image segmentation model. The main purpose of this stage is to let the model first learn the characteristics of benign lesions, laying the foundation for the subsequent introduction of malignant lesion data.

[0046] Next, the proportion of malignant lesion segmentation training data is gradually increased for model training. Specifically, it starts with all benign data, and then gradually introduces some malignant data. In multiple training stages, the proportion of malignant data is gradually increased until all data is finally used. This gradual introduction method helps the model gradually adapt to the characteristics of malignant data and reduce the shock and instability of the model during training.

[0047] When the proportion of malignant data increases to a certain extent (for example, reaches a certain preset proportion or is used entirely), comprehensive training begins using all benign lesion segmentation training data and malignant lesion segmentation training data. The main purpose of this stage is to allow the model to learn the characteristics of both benign and malignant lesions at the same time, thereby improving the segmentation accuracy and generalization ability of the model.

[0048] Finally, on the basis of comprehensive training, all benign lesion segmentation training data and malignant lesion segmentation training data are further enhanced to increase the diversity and richness of the data. Data enhancement techniques may include rotation, scaling, translation, flipping, color conversion, etc. Then, the model is trained to convergence using the data after data enhancement to obtain the first segmentation model. The main purpose of this stage is to further improve the segmentation performance and robustness of the model by increasing the diversity and richness of the data.

[0049] In summary, the above implementation method not only improves the segmentation accuracy and generalization ability of the model, but also reduces the oscillation in the model training process and improves the robustness of the model by gradually increasing the proportion of malignant lesion segmentation training data, comprehensive training and data enhancement processing strategies, providing more comprehensive and effective support for self-training learning and optimization of medical image segmentation.

[0050] Based on the aforementioned scheme, in some implementations of the present application, the intersection difference between the lesion prediction image and the annotated image is calculated to obtain a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area, respectively, including: for each lesion segmentation prediction image in the test set, calculating the intersection with the corresponding annotated image to obtain the common area; subtracting the common area from the annotated image to obtain a pseudo-label image of the missed detection area; and subtracting the common area from the predicted image to obtain a pseudo-label image of the false detection area.

[0051] In the above implementation, for each lesion segmentation prediction image in the test set, an intersection calculation is performed with the corresponding annotated image. The intersection area represents the part that is considered to be a lesion in both the predicted image and the annotated image. After obtaining the intersection area, the intersection area is subtracted from the annotated image, and the result is a pseudo-label image of the missed detection area. The missed detection area represents the part that is considered to be a lesion in the annotated image but is not detected in the predicted image. These areas reflect the shortcomings of the model in detecting lesions and areas that need improvement. Similarly, after obtaining the intersection area, the intersection area is subtracted from the predicted image, and the result is a pseudo-label image of the misdetected area. The misdetected area represents the part that is considered to be a lesion in the predicted image but is not in the annotated image. These areas reflect the misjudgment and noise that may be generated by the model when detecting lesions.

[0052] Among them, the generation of pseudo-label images of missed detection areas and pseudo-label images of false detection areas enables targeted optimization of the model. By adding these pseudo-label images to the training set, the model can learn more features and details, thereby improving segmentation performance and accuracy. And by visualizing pseudo-label images of missed detection and false detection areas, the performance and shortcomings of the model in detecting lesions can be intuitively understood, which improves the interpretability and credibility of the model. In other words, it not only provides a clear direction and goal for the optimization of the model, but also improves the performance of the model, promotes the continuous improvement of the model, and improves the interpretability of the model.

[0053] Based on the aforementioned scheme, in some implementation methods of the present application, the pseudo-label images of missed-detected areas and the pseudo-label images of false-detected areas are added to a training set, and the first segmentation model is trained until convergence, including: iteratively updating the pseudo-label images of missed-detected areas and the pseudo-label images of false-detected areas, and in each iteration, retraining the first segmentation model until convergence using the data in the training data set, and the updated pseudo-label images of missed-detected areas and the pseudo-label images of false-detected areas; wherein, in each iteration, including using independent test set data to evaluate the performance of the model after each iteration, if the accuracy of the model on the test set is improved compared with the previous iteration, retaining the parameters of the current model as the starting point for the next iteration; and using the model after each iteration to calculate the accuracy of the prediction results of the training data set, and updating / not updating the pseudo-label images of missed-detected areas and the pseudo-label images of false-detected areas according to the accuracy of the calculated prediction results.

[0054] In the above implementation, the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area are first iteratively updated. This means that in each iteration, these pseudo-label images are adjusted and updated according to the performance of the current model to better guide the training of the model. In each iteration, the first segmentation model is retrained by using the data in the training data set and the updated pseudo-label images of the missed detection area and the pseudo-label images of the false detection area until the model converges. This step enables the model to gradually learn more features and details, thereby improving the segmentation performance. In addition, after each iteration, by using independent test set data to evaluate the performance of the model, it can be ensured that the model can also show good generalization ability on unseen data. If the accuracy of the model on the test set is improved compared to the previous iteration, then the parameters of the current model are retained as the starting point for the next iteration. At the same time, the model after each iteration is also used to calculate the accuracy of the prediction results of the training data set. This accuracy index reflects the performance of the model on the current training data. According to the accuracy of the calculated prediction results, it is decided whether to update the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area. If the accuracy of the model is improved, it means that the current pseudo-label images have guided the training of the model well, so there may be no need to update it; if the accuracy of the model has not improved or the improvement is not obvious, then it is necessary to update the pseudo-label images to introduce more useful information to further guide the training of the model.

[0055] In summary, the above implementation method not only continuously optimizes the performance of the model but also improves the generalization ability of the model by iteratively updating the pseudo-label images and retraining the model, and adopts an adaptive pseudo-label update strategy to reduce unnecessary computational overhead.

[0056] Based on the aforementioned scheme, in some implementations of the present application, the accuracy of the prediction results of the training data set is calculated using the model after each iteration, and the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area are updated / not updated according to the accuracy of the calculated prediction results, including: if the accuracy of the fused prediction image is higher than the accuracy of the intermediate prediction image, and higher than the accuracy of the fused prediction image retained in the last update, the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area are updated; otherwise, they are not updated.

[0057] In the above implementation, by deciding whether to update the pseudo-label image based on the prediction accuracy, not only unnecessary updates are avoided, but also the quality of the pseudo-labels is ensured, the continuous improvement of the model is promoted, and the training efficiency is improved.

[0058] Based on the aforementioned scheme, in some implementations of the present application, the confidence calculation step of the pseudo-label image of the missed detection area and the pseudo-label image of the misdetected area includes: calculating the pseudo-label effectiveness rate, and calculating the pseudo-label confidence by fusing all the average precisions and pseudo-label effectiveness rates.

[0059] It should be noted that the pseudo-label effectiveness refers to the ratio of the correctly labeled area in the pseudo-label image to the entire pseudo-label image. In order to calculate this ratio, the pseudo-label image (pseudo-label image of missed detection area and pseudo-label image of misdetected area) can be compared with the real labeled image, the number of correctly labeled pixels can be counted, and then divided by the total number of pixels in the pseudo-label image. Among them, the correctly labeled area may include completely correct annotations (i.e., areas that are completely consistent with the real annotations) and partially correct annotations (i.e., areas that partially overlap with the real annotations). When calculating the pseudo-label effectiveness, you can choose whether to count the partially correct annotations into the valid area according to actual needs.

[0060] At the same time, each pixel can be regarded as a sample, and the prediction accuracy of the model on each pixel is calculated, and then the average accuracy of all pixels is calculated. For the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area, their average accuracy can be calculated respectively. This includes comparing the pseudo-label images of the missed detection area and / or the pseudo-label images of the false detection area with the real annotated images, counting whether the prediction result of each pixel is correct, and calculating the average accuracy based on this.

[0061] After obtaining the pseudo-label effectiveness and average precision, they can be combined to calculate the confidence of the pseudo-label. Specifically, the pseudo-label effectiveness can be multiplied by the average precision (or a weighted average of them) to obtain a comprehensive evaluation index, namely the pseudo-label confidence. The higher the confidence of the pseudo-label image in the missed detection area and the pseudo-label image in the false detection area, the higher the reliability and accuracy of the pseudo-label image in the missed detection area and the pseudo-label image in the false detection area. This helps to make more reasonable use of pseudo-label images in the subsequent model training and optimization process.

[0062] Based on the aforementioned scheme, in some implementation methods of the present application, the loss calculation of the training process includes: when training is not performed without adding pseudo-label images of missed detection areas and pseudo-label images of false detection areas, the SoftMax cross entropy loss is calculated for the intermediate prediction features calculated by the model; when training is performed with pseudo-label images of missed detection areas and pseudo-label images of false detection areas, the SoftMax cross entropy loss is calculated for the intermediate prediction features calculated by the model, and the SoftMax cross entropy loss and consistency loss are calculated for the fused prediction features, and the SoftMax cross entropy loss or L1 loss (i.e., absolute error loss) is calculated for the prediction features of the missed detection areas and the prediction features of the false detection areas.

[0063] In the above implementation, considering that in the early stage of training, when pseudo-label images of missed detection areas and false detection areas have not yet been added, the main focus should be on the model's basic segmentation capabilities for benign and malignant lesions. At this time, the SoftMax cross entropy loss is calculated for the intermediate prediction features calculated by the model. SoftMax cross entropy loss is a commonly used loss function in classification tasks, which can measure the difference between the probability distribution predicted by the model and the probability distribution of the true label.

[0064] When pseudo-label images of missed detection areas and pseudo-label images of false detection areas are added for training, the above implementation introduces more supervisory information to guide the training of the model. As when pseudo-label images are not added, the SoftMax cross entropy loss is still calculated for the intermediate prediction features of the model to keep the model's focus on the basic segmentation task. At the same time, the SoftMax cross entropy loss is calculated for these fused prediction features to evaluate the segmentation performance of the model after the fusion information. And the consistency loss is calculated to encourage the model to produce consistent prediction results under different inputs. For the prediction features of missed detection areas and the prediction features of false detection areas, the SoftMax cross entropy loss or L1 loss is calculated respectively. The SoftMax cross entropy loss is used to evaluate the classification performance of the model on these areas, while the L1 loss (i.e., absolute error loss) is used to measure the absolute difference between the model prediction results and the true annotations.

[0065] In summary, in the above implementation, by introducing multiple loss functions to comprehensively evaluate and optimize the performance of the model, the model can perform well in many aspects (such as basic segmentation ability, fusion information processing ability, and the ability to identify missed and falsely detected areas, etc.), which helps to improve the overall performance and accuracy of the model.

[0066] In order to make the present application more intuitive for those skilled in the art, a specific example will be used to illustrate it. It should be noted that the embodiments of the present application can be applied to the training of various medical image segmentation models. For the convenience of description and understanding, the subsequent examples are described by taking the segmentation of breast tumor lesions as an example, but those skilled in the art can extend it to the training of other medical image segmentation models, which is not limited here.

[0067] First, the breast tumor segmentation training set (i.e., the original training data set) is defined as Where N s represents the number of breast ultrasound images in the training set, x i and q i They represent the i-th breast ultrasound image and the corresponding breast tumor segmentation annotation image (Ground-Truth). Define q i The size of is H×W, where H and W represent the length and width of the labeled image, respectively. Then, the breast tumor segmentation training set is further split into a benign breast tumor segmentation training data set and a malignant breast tumor segmentation training data set (i.e., benign lesion segmentation training data and malignant lesion segmentation training data), respectively denoted as and D s1 represents the benign breast tumor segmentation training dataset, D s2 represents the malignant breast tumor segmentation training dataset, N s1 and N s2 Respectively represent the number of breast ultrasound images in the benign breast tumor segmentation training dataset and the malignant breast tumor segmentation training dataset. Similarly, define is the breast tumor segmentation test set (i.e., the original test data set), N l represents the number of breast ultrasound images in the training set. Define f θ :x i →q i , to optimize the accuracy of the model by training a breast tumor segmentation network.

[0068] Self-training course learning plan:

[0069] Considering that self-training with a whole batch of pseudo-labels will eventually produce noisy labels, affecting the effectiveness of the model. To alleviate this situation, this example proposes a self-training curriculum learning framework for breast tumor segmentation, which encourages learning effective information from high-confidence pseudo-labels. Figure 2 As shown in the figure, the proposed self-training curriculum learning framework includes curriculum learning, feature fusion, pseudo label generation, pseudo label update, pseudo label confidence calculation and loss calculation.

[0070] In order to better identify breast tumors of different forms, a curriculum learning scheme is first introduced at the data training level and pseudo-label optimization level. At the same time, in order to prevent the generated pseudo-labels from overfitting, the breast tumor segmentation training set is and breast tumor segmentation test set Expand into three sets of training sets and test sets for iterative training.

[0071] First, the breast tumor segmentation training set Divide into two data sets according to the number of people, denoted as and Then, fusion and data to form a new training set Fusion and data to form a new training set Thus, three sets of training sets and test sets are formed, namely the first set of training sets The first test set The second training set The second test set The third training set The third test set

[0072] First, at the data training level, consider training the initial breast tumor segmentation model as the breast tumor teacher segmentation network through a course learning program, which can be implemented in four steps. Taking the first set of data as an example, define Represents the model learned by the first set of breast tumor segmentation data at the tth update Representation Model Model parameters updated for the tth time:

[0073] The first step is to use the benign breast tumor segmentation training dataset D s1 As training data, use Swin-Unet as the backbone network and model Train until convergence.

[0074] The second step considers the data D of benign tumors s1 and malignant tumor data s2 Fusion of benign tumor data D s1 and some data on malignant tumors (D s2In the selection of malignant tumor data, the curriculum learning scheme is considered, and the malignant tumor data is gradually increased in multiple training stages. s2 Randomly select 10% of the data and the benign tumor data D s1 Then, in each epoch of training, the proportion of malignant breast tumor data and all benign breast tumor data randomly extracted from the malignant breast tumor database to form a training set is increased by 1% compared with the previous epoch, until the proportion of malignant breast tumors is increased to 100%, and the model is gradually trained.

[0075] The third step uses all the benign tumor data D in the training set s1 and malignant tumor data s2 For training, use the breast tumor segmentation training set D s Breast tumor segmentation model Train until convergence.

[0076] The fourth step is to train the breast tumor database using data enhancement methods, including rotation transformation, flip transformation, scaling transformation, scale transformation, noise perturbation and color change. The initial sample size is expanded to N s Twice, every 10 epochs, increase by 2N s The number of samples is increased until the number of samples with data enhancement in each epoch reaches 12N. s Finally, we use 12 times the number of data augmentation images to train the model. Continue training until the model converges. During each epoch of training, the parameters of data augmentation are randomly adjusted, that is, the samples of data augmentation are different in each iterative training.

[0077] Waiting for model After convergence, we use the breast tumor segmentation test set D l As input, we pass Compute lesion prediction image set And save it to the database, and calculate the corresponding Dice precision set and retain, where r i = Dice(p i ,q i ).

[0078] Similarly, the second group of breast tumor segmentation training set D lu And the third group of breast tumor segmentation training set D lv Train and get the model trained by the second set of data The model trained with the second set of data Then use the model and Model The second group of breast tumor segmentation test set D v And the third group of breast tumor segmentation test set D u The images are tested to obtain the lesion prediction sets of the second test set. And the lesion prediction set of the third test set Finally, P u and P v The corresponding Dice accuracy is calculated based on the lesion image, denoted as and

[0079] Pseudo label calculation:

[0080] Then use P l , P u and P v Calculate the set of pseudo-label images of missed detection areas and false detection areas of the three test sets and

[0081] For P l , P u and P v Each breast tumor segmentation prediction image p i , and compare it with the corresponding annotation image q i Calculate the intersection and get their common area c i Then, use q i Subtract c i , you can get the pseudo-label image of the missed area Similarly, p i Subtract c i , then we can get the false detection area pseudo label image Thus, we can get P l , P u and P v The corresponding set Q of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas l , Q u and Q v .

[0082] Because D l , D u and D v The corresponding breast ultrasound images cover all the data of the training set and the test set, so P l , P u and P v The calculated Q l , Q uand Q v It can also cover three sets of training sets. Therefore, and The pseudo-label images of the missed detection area and the false detection area are added to the training set D s , D lu and D lv In the above example, three new training sets are formed, which are respectively denoted as and Then use and training data, 12 times the data enhancement, and continue to improve the model Model and Model Continue training until the model converges.

[0083] Self-training learning:

[0084] Finally, by incorporating a self-training learning scheme, the pseudo-label images of missed detection areas and false detection areas are continuously optimized iteratively, thereby improving the performance of the breast tumor segmentation network. In the self-training course learning process, the main steps include feature fusion, pseudo-label update, pseudo-label confidence calculation, and loss calculation, which are introduced as follows:

[0085] Feature Fusion:

[0086] like Figure 3 As shown, the lesion area prediction feature F in the network p , Prediction feature of missed area F m And the predicted features of the misdetected area F f As input, it is input to the feature fusion module. First, F p 、F m and F f The three feature layers are cascaded to obtain feature F c . Then feature F c It is followed by three fully connected layers, the output channels of which are 4096, 512, and 3, respectively, to output 4096-dimensional, 512-dimensional, and 3-dimensional features. At the same time, the first two fully connected layers are followed by Leak ReLu layers to enhance the nonlinear ability of the network. Then, after the 3-dimensional features, a Softmax layer is added to normalize the interval of the feature values ​​to between 0 and 1, and a 3-dimensional weight array [ω1, ω2, ω3] is obtained. Finally, F p 、F m and F f Perform weighted processing to obtain fusion prediction features The formula is as follows:

[0087]

[0088] That is, the weighted lesion region features are added to the weighted missed detection region features and then subtracted from the weighted misdetected region features to obtain the missed detection region prediction features.

[0089] Pseudo label update:

[0090] Pseudo-label images of missed areas through self-training And false detection area pseudo label image Carry out continuous iterative updates.

[0091] The specific steps to calculate and update pseudo labels are as follows:

[0092] Take the first set of training data As input, after the model converges, use the model after each epoch For the test set D l If in the test set D l Model on Comparison The overall accuracy of the model is improved, so the model is retained. Model In the test set D l An iterative update is considered an improvement in accuracy.

[0093] If you save the updated new model Then we use the new model To pseudo-label Update. and Respectively represent through x i The lesion area prediction features and fusion prediction features are calculated. and Perform Argmax operation, which means passing x i The calculated intermediate prediction image p ti And fusion prediction image Then calculate P ti and The accuracy of ti and Denoted as r ti = Dice(p ti ,q i )and And use r ti and The previous model remains Pseudo-labeled images of missed areas for breast tumor segmentation tasks And false detection area pseudo label image to update.

[0094] make and Respectively represent the image through x i The prediction features of missed detection area and false detection area are calculated. and Perform Argmax operation to obtain the missed detection area prediction image of breast tumor segmentation with category values ​​0 and 1 And the predicted image of the false detection area Then calculate separately and With q i The intersection of and q i Public areas of c i1 as well as and q i Public areas of c i2 Then, let Represents the input image x i By Model Calculate the pseudo label of the missed area make Indicates that by x i Calculate the false positive area pseudo label

[0095] Assume that the fusion prediction image Accuracy Higher than the intermediate prediction image P ti The accuracy of ti , then it is determined that the prediction features of the missed detection area are generated and false positive region prediction features is effective. At the same time, if the fusion prediction image Accuracy Higher than the accuracy of the fused prediction image retained in the database at the last update When , the pseudo-label image of the missed area is considered And false detection area pseudo label image to update.

[0096] The tth updated and It can be defined as:

[0097]

[0098] That is, for the input image x i In terms of Accuracy Higher than P ti The accuracy of tiThe accuracy of the fusion prediction results retained in the database at the last update , then the pseudo-label image of the missed area And false detection area pseudo label image to update.

[0099] At the same time, update and After that, the tth retained To update, define it as:

[0100]

[0101] That is, if x i Pseudo-label image of the missed area And false detection area pseudo label image When updating, the Dice accuracy of the updated fusion prediction features is also retained accordingly. Otherwise, the Dice precision stored in the database during the last update is retained. When performing the first self-training update, let Use retention before self-training learning r i To calculate the Dice accuracy update.

[0102] Through the above operations, we can get the first set of training set data Iteratively updated model , the first test set D l In the model The set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas calculated after the t-th iteration update and D l In the model The Dice precision set calculated after the tth iteration update

[0103] Similarly, using the second set of training data The iteratively updated model can be calculated The second test set Dv is used in the model The set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas calculated after the t-th iteration update and D v In the model The Dice precision set calculated after the tth iteration update Using the third set of training data The iteratively updated model can be calculated The third test set D u On the model The set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas calculated after the t-th iteration update and D u In the model The Dice precision set calculated after the tth iteration update

[0104] The process is consistent with the pseudo-label calculation process, because D l , D u and D v The corresponding breast ultrasound images cover all the data of the training set and the test set, so the updated Q l , Q u and Q v It can also cover three training sets.

[0105] Fusion Q u and Q v The data can be obtained The corresponding set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas is denoted as Fusion Q l and Q u data, that is, The corresponding set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas is denoted as Fusion Q l and Q v data, that is, The corresponding set of pseudo-label images of missed detection areas and pseudo-label images of misdetected areas is denoted as Then will and The pseudo-label images of the missed detection areas and the pseudo-label images of the misdetected areas in the image are replaced by Q s , Q lu and Q lv The data of the training set after the model is updated for the tth time can be obtained. and

[0106] Pseudo label confidence calculation:

[0107] Because the pseudo-label images of the missed detection area and the false detection area of ​​the breast tumor segmentation task are generated, there is no corresponding supervisory information to calculate the accuracy. However, the pseudo-labels of the breast tumor segmentation task are used to assist the estimation of the breast tumor segmentation task. To this end, the label information of the breast tumor segmentation task is used to calculate the confidence of the pseudo-label images of the missed detection area and the false detection area.

[0108] Let x il Represents x iFor the image after data enhancement, the pseudo-label efficiency is first calculated, which is defined as:

[0109]

[0110] Where L = 12, indicating the number of data enhancements. Let p til and Indicates that by x il The calculated intermediate prediction image and fused prediction image are and r til Respectively represent p til and Dice accuracy. ρ is a discriminant, when , ρ=1, otherwise ρ=0.

[0111] Then, by combining all The average precision and pseudo-label efficiency are used to calculate the pseudo-label confidence, which is defined as:

[0112]

[0113] Then the average confidence of the pseudo-label after the t-th model update is calculated, which is defined as:

[0114]

[0115] Where N represents the number of training samples, and the values ​​of N in the three training sets are equal to N s 、N lu and N lv .

[0116] Finally, for x i The pseudo label confidence when , then it is considered that the tth updated and The confidence level is high; when , then it is considered that the tth updated and The confidence level is low.

[0117] Loss calculation:

[0118] The proposed self-training curriculum learning framework uses different losses during training and is implemented in two stages.

[0119] First, when the pseudo-label images of missed detection areas and false detection areas are not added for training, the SoftMax cross entropy loss is calculated for the intermediate prediction features.

[0120] Second, when adding pseudo-label images of missed detection areas and pseudo-label images of false detection areas for training, in addition to calculating the SoftMax cross entropy loss for the intermediate prediction features, the SoftMax cross entropy loss and consistency loss are also calculated for the fused prediction features. At the same time, the SoftMax cross entropy loss or L1 loss is calculated for the prediction features of missed detection areas and false detection areas.

[0121] First, when training without adding pseudo-label images of missed detection areas and pseudo-label images of misdetected areas, directly use To calculate the SoftMax cross entropy loss function, it is defined as:

[0122]

[0123] in, It is right i The result of one-hot encoding is: Yes Calculate the result of the SoftMax operation, where B represents the number of samples in the same Batch.

[0124] After adding the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area for model training, continue to fusion prediction features Calculate the cross entropy loss, defined as:

[0125]

[0126] in, Express Calculate the result of the SoftMax operation.

[0127] Considering the fusion prediction image There may be discontinuities in the predictions within a region, so consistency loss is used to optimize the model training. and q i Multiplying by a 3×3 convolution [[0,-1,0],[-1,4,-1],[0,-1,0]] respectively, and then calculating the L2 distance loss, we can get the consistency loss, which is defined as:

[0128]

[0129] in, Represents a convolution operation.

[0130] Next, consider calculating the loss of the pseudo-label images of the missed detection area and the false detection area. When the confidence of the pseudo-label images of the missed detection area and the false detection area is high, choose to use the SoftMax cross entropy loss function for optimization. When the confidence of the pseudo-label images of the missed detection area and the false detection area is low, choose to use the L1 distance loss for optimization, which is defined as:

[0131]

[0132] in, and are respectively and The result of one-hot encoding. and are respectively and Calculate the result of the SoftMax operation.

[0133] Finally, when calculating the total loss, the consistency loss of the fused prediction features and the pseudo-label loss are considered as auxiliary losses, and the total loss updated using the t-th pseudo-label is defined as:

[0134]

[0135] Exemplarily, the weight values ​​of γ and μ may be set to 0.3.

[0136] Example 2

[0137] See also Figure 4 , the embodiment of the present application provides a self-training learning system for medical image segmentation, which includes:

[0138] The data acquisition module is configured to: acquire an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data. The data preprocessing module is configured to: divide the original training data set into multiple data sets, and perform iterative training in combination with the original test data set to form multiple groups of training data sets and test data sets. The model training module is configured to: respectively use each group of training data sets and test data sets formed by iterative training to execute a training process, and the training process includes: gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model; using the first segmentation model to predict the test data set to obtain the corresponding lesion prediction image; by calculating the intersection difference between the lesion prediction image and the annotated image, respectively obtain a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area; adding the obtained pseudo-label image of the missed detection area and the pseudo-label image of the false detection area to the training set, and training the first segmentation model until convergence.

[0139] For the specific implementation process of the above system, please refer to the self-training learning method for medical image segmentation provided in Example 1, which will not be repeated here.

[0140] Example 3

[0141] See also Figure 5 , the embodiment of the present application provides an electronic device, which includes at least one processor 201 and at least one memory 202; wherein the processor 201 and the memory 202 are directly connected to each other, or communicate with each other through a communication interface 203, or are electrically connected through one or more communication buses or signal lines to achieve data transmission or interaction; the memory 202 stores program instructions that can be executed by the processor 201, and the processor 201 calls the program instructions to execute a self-training learning method for medical image segmentation. For example, the following is implemented:

[0142] An original training data set and an original test data set are obtained; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data. The original training data set is evenly divided into multiple data sets, and iterative training is performed in combination with the original test data set to form multiple groups of training data sets and test data sets. The training process is performed using each group of training data sets and test data sets formed by iterative training, and the training process includes: gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model; using the first segmentation model to predict the test data set to obtain a corresponding lesion prediction image; by calculating the intersection difference between the lesion prediction image and the annotated image, a pseudo-label image of the missed detection area and a pseudo-label image of the false detection area are obtained respectively; the obtained pseudo-label image of the missed detection area and the pseudo-label image of the false detection area are added to the training set, and the first segmentation model is trained until convergence.

[0143] Among them, the memory 202 can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc.

[0144] The processor 201 may be an integrated circuit chip with signal processing capability. The processor 201 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0145] Understandably, Figure 5 The structure shown is for illustration only. The electronic device may also include Figure 5 More or fewer components as shown, or with Figure 5 Different configurations are shown. Figure 5 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0146] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A self-training learning method for medical image segmentation, characterized in that: The following steps are involved: Acquire an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data; The original training data set is divided into multiple data sets, and iterative training is performed on the original test data set to form multiple sets of training data sets and test data sets; The training data sets and the test data sets formed by iterative training are respectively used to execute a training process, wherein the training process includes: The proportion of malignant lesion segmentation training data is gradually increased through multiple training stages, and the initial medical image segmentation model is trained until convergence to obtain a first segmentation model; Using the first segmentation model to predict the test data set, obtaining a corresponding lesion prediction image; By calculating the intersection difference between the lesion prediction image and the annotated image, the pseudo-label image of the missed detection area and the pseudo-label image of the misdetected area are obtained respectively; The obtained pseudo-label images of the missed detection area and the pseudo-label images of the misdetected area are added to the training set, and the first segmentation model is trained until convergence.

2. The method according to claim 1, characterized in that The original training data set is divided into multiple data sets, and iterative training is performed on the original test data set to form multiple sets of training data sets and test data sets, including: The original training data set is divided equally into a first training data set and a second training data set according to the number of people; The original training data set and the original test data set are used as the first set of training data sets and test data sets; The original test data set and the first training data set are combined to obtain a new test data set, and the second training data set is used as the corresponding test data set to obtain a second set of training data sets and a test data set; The original test data set and the second training data set are fused to obtain a new test data set, and the first training data set is used as the corresponding test data set to obtain a third set of training data sets and test data sets.

3. The method according to claim 1, characterized in that The method of gradually increasing the proportion of malignant lesion segmentation training data through multiple training stages, training the initial medical image segmentation model until convergence, and obtaining a first segmentation model includes: Using the benign lesion segmentation training data in the training data set of the current group to train the original medical image segmentation model until convergence; The model is trained using all benign lesion segmentation training data and part of the malignant lesion segmentation training data, and the malignant lesion segmentation training data is gradually increased in multiple training stages until all of it is used; Use all benign lesion segmentation training data and malignant lesion segmentation training data to train the model until convergence; Data enhancement processing is performed on all benign lesion segmentation training data and malignant lesion segmentation training data, and the model is trained until convergence using the data after data enhancement processing to obtain the first segmentation model.

4. The method according to claim 1, characterized in that: The method of calculating the intersection difference between the lesion prediction image and the annotated image to obtain a pseudo-label image of the missed detection area and a pseudo-label image of the misdetected area respectively includes: For each lesion segmentation prediction image in the test set, calculate the intersection with the corresponding annotated image to obtain the common area; Subtract the common area from the annotated image to obtain a pseudo-label image of the missed area; and Subtract the common area from the predicted image to obtain the pseudo-label image of the misdetected area.

5. The method according to claim 1, characterized in that The method of adding the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area to the training set and training the first segmentation model until convergence includes: Iteratively updating the pseudo-label images of the missed detection area and the pseudo-label images of the false detection area, and in each iteration, retraining the first segmentation model until convergence using the data in the training data set and the updated pseudo-label images of the missed detection area and the pseudo-label images of the false detection area; In each iteration, the performance of the model after each iteration is evaluated using independent test set data. If the accuracy of the model on the test set is improved compared to the previous iteration, the parameters of the current model are retained as the starting point for the next iteration; and The model after each iteration is used to calculate the accuracy of the prediction results of the training data set, and the pseudo-label images of the missed detection area and the pseudo-label images of the misdetected area are updated / not updated according to the accuracy of the calculated prediction results.

6. The method according to claim 5, characterized in that The method of calculating the accuracy of the prediction results of the training data set using the model after each iteration, and updating / not updating the pseudo-label images of the missed detection area and the pseudo-label images of the misdetected area according to the accuracy of the calculated prediction results, includes: If the accuracy of the fused prediction image is higher than the accuracy of the intermediate prediction image and higher than the accuracy of the fused prediction image retained during the last update, the pseudo-label image of the missed detection area and the pseudo-label image of the misdetected area are updated; otherwise, they are not updated.

7. The method according to claim 1, characterized in that The confidence calculation steps of the pseudo-label image of the missed detection area and the pseudo-label image of the false detection area include: calculating the pseudo-label effectiveness rate, and calculating the pseudo-label confidence by fusing all the average precisions and pseudo-label effectiveness rates.

8. The method according to claim 1, characterized in that The loss calculation of the training process includes: When training without adding pseudo-label images of missed detection areas and pseudo-label images of misdetected areas, the SoftMax cross entropy loss is calculated for the intermediate prediction features calculated by the model; When adding pseudo-label images of missed detection areas and pseudo-label images of false detection areas for training, the SoftMax cross entropy loss is calculated for the intermediate prediction features calculated by the model, and the SoftMax cross entropy loss and consistency loss are calculated for the fused prediction features, as well as the SoftMax cross entropy loss or L1 loss for the prediction features of missed detection areas and false detection areas.

9. A self-training learning system for medical image segmentation, characterized in that: include: The data acquisition module is configured to: acquire an original training data set and an original test data set; the original training data set includes benign lesion segmentation training data and malignant lesion segmentation training data; The data preprocessing module is configured to: divide the original training data set into multiple data sets, and perform iterative training on the original test data set to form multiple sets of training data sets and test data sets; The model training module is configured to: respectively use each set of training data sets and test data sets formed by iterative training to execute a training process, wherein the training process includes: The proportion of malignant lesion segmentation training data is gradually increased through multiple training stages, and the initial medical image segmentation model is trained until convergence to obtain a first segmentation model; Using the first segmentation model to predict the test data set, obtaining a corresponding lesion prediction image; By calculating the intersection difference between the lesion prediction image and the annotated image, the pseudo-label image of the missed detection area and the pseudo-label image of the misdetected area are obtained respectively; The obtained pseudo-label images of the missed detection area and the pseudo-label images of the misdetected area are added to the training set, and the first segmentation model is trained until convergence.

10. An electronic device, characterized in that: include: A memory for storing one or more programs; processor; When the one or more programs are executed by the processor, the method according to any one of claims 1 to 8 is implemented.

Images