Cancer histological subtype classification model training method and storage medium

Abstract

The invention provides a cancer histological subtype classification model training method and a storage medium, and the method comprises the steps: recognizing a tumor image region in an HE staining section image, and dividing a plurality of non-overlapping sub-section images in the tumor image region; for each sub-slice image corresponding to each HE dyeing slice image in the sample set, inputting the sub-slice image into a feature extraction network to obtain a basic feature vector of the sub-slice image; inputting the basic feature vectors of all sub-slice images of the HE staining slice image into an attention network and a mixed density network which are connected with each other to construct an enhanced feature vector of the HE staining slice image; inputting the enhanced feature vector of the HE staining slice image into a classifier to output a cancer histological subtype prediction result of the HE staining slice image; and adjusting the hyper-parameters to generate a trained cancer histological subtype classification model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a training method of a cancer histological subtype classification model and a storage medium. BACKGROUND

[0002] With the wide application of deep learning technology in the field of medical image analysis, deep learning models such as convolutional neural network (CNN) and recurrent neural network (RNN) have been widely used in the classification task of pathological images. In the prior art, the classification accuracy is usually improved by optimizing the network structure or feature aggregation strategy, but there is a lack of subtype classification optimization for subtle histological differences. For example, when distinguishing different molecular subtypes of hepatocellular carcinoma, the existing method may ignore some low-probability but discriminative pathological features, thereby affecting the accuracy of classification. SUMMARY

[0003] The purpose of the embodiments of the present application is to provide a training method of a cancer histological subtype classification model and a storage medium, so as to improve the classification accuracy of the cancer subtype classification model.

[0004] In a first aspect, the present application provides a training method of a cancer histological subtype classification model, which comprises: For each HE-stained section image in the sample set, identifying a tumor image region in the HE-stained section image, and dividing a plurality of non-overlapping sub-section images in the tumor image region; For each sub-section image corresponding to each HE-stained section image in the sample set, inputting the sub-section image into a feature extraction network to obtain a basic feature vector of the sub-section image; For each HE-stained section image in the sample set, inputting the basic feature vectors of all sub-section images of the HE-stained section image into a connected attention network and a hybrid density network to construct an enhanced feature vector of the HE-stained section image; For each HE-stained section image in the sample set, inputting the enhanced feature vector of the HE-stained section image into a classifier to output a cancer histological subtype prediction result of the HE-stained section image; For each HE-stained section image in the sample set, adjusting the hyperparameters based on the comprehensive loss value corresponding to the attention network, the hybrid density network and the classifier to generate a trained cancer histological subtype classification model.

[0005] In an optional embodiment, the enhanced feature vector of the target HE-stained section image is constructed by: The basic feature vectors of all sub-slice images of the target HE-stained slice image are input into the attention network to obtain the attention score and optimized feature vector of each sub-slice image output by the attention network; The attention scores and optimized feature vectors of all sub-slices of the target HE-stained slice image are input into a hybrid density network to obtain the enhanced feature vector of the target HE-stained slice image output by the hybrid density network.

[0006] In an optional implementation, the step of the attention network outputting the optimized feature vector for each sub-slice image specifically includes: The basic feature vectors of all sub-slice images of the target HE-stained slice image are input into the attention network to calculate the attention score of each sub-slice image; Based on the attention scores of all sub-slice images of the target HE staining slice image, the weight coefficient value corresponding to each sub-slice image is determined; For each sub-slice of the target HE-stained section image, the optimized feature vector of the sub-slice image is calculated based on the basic feature vector of the sub-slice image and the corresponding weight coefficient value.

[0007] In an optional implementation, the step of the hybrid density network outputting the enhanced feature vector of the target HE-stained slice image specifically includes: The attention scores and optimized feature vectors of all sub-slice images of the target HE staining slice image are input into the mixing density network to calculate the mixing coefficients, mean and standard deviation of all sub-slice images. For each sub-slice image of the target HE staining section image, the cluster assignment probability value of the sub-slice image is calculated based on the attention score, mixing coefficient value, mean and standard deviation value of the sub-slice image; Based on the class assignment probability value and mean of each sub-slice image, the regression prediction value of that sub-slice image is calculated; The enhanced feature vector is obtained by weighting the optimized feature vectors and regression predictions of all sub-slice images of the target HE-stained slice image.

[0008] In an optional implementation, the cluster assignment probability value is calculated using the following formula: ; in, For the first The sub-slice image belongs to the first The probability values ​​assigned to each cluster. The first output of the attention network Attention scores for individual slice images, The output of the hybrid density network The mean coefficients corresponding to each sub-slice image. The output of the hybrid density network The square of the standard deviation coefficients corresponding to each sub-slice image The output of the hybrid density network The mixing coefficients corresponding to each sub-slice image; Furthermore, the regression prediction value is calculated using the following formula: ; in, For the first The regression prediction value corresponding to each sub-slice image.

[0009] In an optional implementation, the loss value is calculated using the following formula: ; ; ; ; in, For classifying losses, For attention regression loss, For the mixed density loss, For the actual results of cancer histological subtypes, The results are for predicting cancer histological subtypes.

[0010] In an optional implementation, the weight coefficient value corresponding to the sub-slice image is determined by the following formula: ; in, For the first Attention scores corresponding to each sub-slice image.

[0011] In an optional implementation, the HE-stained slide image to be classified is input into a pre-trained cancer histology subtype classification model to obtain the cancer histology subtype classification result output by the cancer histology subtype classification model. The cancer histology subtype classification result is used to indicate whether the HE-stained slide image to be classified belongs to the target cancer histology subtype.

[0012] In a second aspect, the present invention provides an electronic device, comprising: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of a training method for a classification model of cancer histological subtypes as described in any of the foregoing embodiments.

[0013] In a third aspect, the present application provides a computer readable storage medium, which stores a computer program. When the computer program is run by a processor, the steps of the training method of the histological subtype classification model of cancer are performed. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments of the present application will be briefly introduced as follows. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0015] Figure 1 A flowchart of the training method of the histological subtype classification model of cancer provided by the embodiments of the present application; Figure 2 A structural schematic diagram of the histological subtype classification model of cancer provided by the embodiments of the present application; Figure 3 A structural schematic diagram of the electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0016] The technical solutions of the embodiments of the present application will be described below in combination with the drawings in the embodiments of the present application.

[0017] Figure 1 A flowchart of the training method of the histological subtype classification model of cancer provided by the embodiments of the present application, Figure 2 A structural schematic diagram of the histological subtype classification model of cancer provided by the embodiments of the present application. As shown in Figure 1 and Figure 2 As shown in the embodiments of the present application, a training method of a histological subtype classification model of cancer is provided, which comprises: S1, for each HE staining (hematoxylin-eosin staining) section image in the sample set, identifying a tumor image region in the HE staining section image, and dividing a plurality of non-overlapping sub-section images in the tumor image region.

[0018] Here, taking the histological subtype classification diagnosis of a tumor (such as hepatocellular carcinoma) with strong intratumoral heterogeneity as an example, a plurality of whole field digital images (Whole Slide Image, WSI) can be collected, and after preprocessing, the tumor image region is identified by mask separation. In this way, the focus of subsequent analysis can be concentrated on the diagnosis related region.

[0019] In the tumor image region, sub-slice images of 256x256 pixels are divided into non-overlapping patches, denoted as patch, as independent instances in subsequent multiple instance learning, avoiding noise introduced by stroma or normal liver tissue.

[0020] S2, for each sub-slice image corresponding to each HE-stained slice image in the sample set, input the sub-slice image into the feature extraction network to obtain the basic feature vector of the sub-slice image.

[0021] In step S2, ResNet-50 network can be used as a feature extraction network to extract features from each patch. ResNet-50 network can be pre-trained on ImageNet to be applied to histological feature extraction task. Each 256x256 patch is input into the convolutional network of ResNet-50, and a 1024-dimensional feature vector is output.

[0022] S3, for each HE-stained slice image in the sample set, input the basic feature vectors of all sub-slice images of the HE-stained slice image into the interconnected attention network and hybrid density network to construct the enhanced feature vector of the HE-stained slice image.

[0023] Here, the cancer subtype classification problem is regarded as a weakly supervised multiple instance learning (MIL) problem. Each WSI is regarded as a bag containing N patch instances. The basic feature vector of each patch is obtained by ResNet-50 extraction and is mapped to a latent representation space. In order to emphasize the most diagnostic patches, an attention mechanism can be used. The features of each patch are mapped to the latent representation space and activated by the fully connected layer. Then, a softmax function is used to calculate an attention score for each patch, and the basic feature vector of the patch is corrected according to the attention weight. In this way, the attention mechanism helps the model focus on patches containing morphological features with discriminative characteristics.

[0024] In step S3, the enhanced feature vector of the target HE-stained slice image can be constructed in the following way: The basic feature vectors of all sub-slice images of the target HE-stained slice image are input into the attention network to obtain the attention score and optimized feature vector of each sub-slice image output by the attention network. The attention score and optimized feature vector of all sub-slice images of the target HE-stained slice image are input into the hybrid density network to obtain the enhanced feature vector of the target HE-stained slice image output by the hybrid density network.

[0025] Specifically, the steps by which the attention network outputs the optimized feature vector for each sub-slice image include: The basic feature vectors of all sub-slices of the target HE-stained section image are input into an attention network to calculate the attention score for each sub-slice. Based on the attention scores of all sub-slices of the target HE-stained section image, the weight coefficient value corresponding to each sub-slice is determined. For each sub-slice of the target HE-stained section image, the optimized feature vector of that sub-slice image is calculated based on its basic feature vector and corresponding weight coefficient value.

[0026] For each target HE-stained slice image, there can be N sub-slice images. The attention network includes multiple linear layers, scaled dot-product attention layers, and fully connected layers (concat). The attention layer can compute the... Attention score of sub-slice images . The weight coefficient values ​​corresponding to the sub-slice images are determined using the following formula: ; in, It is an exponential function. For the first Attention scores corresponding to each sub-slice image .

[0027] The corresponding weight coefficient value is determined by the proportion of attention scores of the sub-slice image among all sub-slice images.

[0028] Use weighted coefficient values ​​to adjust the underlying feature vectors corresponding to the sub-slice images. Optimize and generate corresponding optimized feature vectors. .

[0029] A hybrid density network is used to fit the attention score for each sub-slice image.

[0030] Specifically, the steps of the hybrid density network outputting the enhanced feature vector of the target HE-stained slice image include: The attention score and the optimized feature vector of each sub-slice image of the target HE-stained slice image are input into the hybrid density network to calculate the hybrid coefficient value, the mean value and the standard deviation value corresponding to each sub-slice image. For each sub-slice image of the target HE-stained slice image, the cluster assignment probability value of the sub-slice image is calculated based on the attention score, the hybrid coefficient value, the mean value and the standard deviation value of the sub-slice image. The regression prediction value of each sub-slice image is calculated based on the cluster assignment probability value and the mean value of each sub-slice image. The enhanced feature vector is obtained by weighting based on the optimized feature vector and the regression prediction value of all sub-slice images of the target HE-stained slice image.

[0031] In an implementable embodiment, the hybrid density network adopts mixed components to regress the optimized feature vector of the patch, and in a specific example may be 3, the parameters (mean coefficient, standard deviation coefficient, hybrid coefficient) of the Gaussian distribution are output to fit the attention score of the patch, the regression prediction value of each sub-slice image is output to form a regression prediction value matrix of .

[0032] Specifically, in the hybrid density network, the optimized feature vector of the patch can be mapped to the hidden space by a fully connected layer.

[0033] The weight coefficient , represents the weight of the th mixed component of the th sub-slice image, and after the Softmax activation function, the weight coefficient of each mixed component corresponding to each sub-slice image is output, and a matrix of is formed for each target HE-stained slice image.

[0034] The mean coefficient = , represents the mean of the th Gaussian distribution corresponding to the th sub-slice image. The mean coefficient of each sub-slice image is output, and a matrix of is formed for each target HE-stained slice image.

[0035] The standard deviation coefficient , represents the standard deviation of the th Gaussian distribution corresponding to the th sub-slice image. The standard deviation coefficient of each sub-slice image is output, and a matrix of ​​The coefficient of variation is 1, and the size of each target HE-stained slice image is 1. The matrix.

[0036] The hybrid density network calculates cluster assignment probabilities based on the fitting results and outputs them as a cluster assignment probability matrix, the size of which is... Specifically, the cluster assignment probability value can be calculated using the following formula: ; in, For the first The sub-slice image belongs to the first The probability values ​​assigned to each cluster. The first output of the attention network Attention scores for individual slice images, The output of the hybrid density network The mean coefficients corresponding to each sub-slice image. The output of the hybrid density network The square of the standard deviation coefficients corresponding to each sub-slice image The output of the hybrid density network The mixing coefficients corresponding to each sub-slice image; All cluster assignment probability values ​​can be normalized before use.

[0037] The regression predictions are used to re-aggregate the weighted features, resulting in a global, slide-level enhanced feature vector. The regression predictions can be calculated using the following formula: ; in, For the first The regression prediction value corresponding to each sub-slice image.

[0038] Finally, the optimized feature vectors of all sub-slice images of the target HE-stained section image are weighted and summed using the regression predicted values ​​to obtain the enhanced feature vector. This enhancement of feature vectors allows the model to capture more complex subtype features.

[0039] S4. For each HE-stained slide image in the sample set, input the enhanced feature vector of the HE-stained slide image into the classifier to output the cancer histological subtype prediction result of the HE-stained slide image.

[0040] The enhanced feature vector is passed through a fully connected layer and a sigmoid activation function to predict the final subtype classification result. In S4, the classifier can be used for binary classification prediction, which is predicted by the enhanced feature vector of the target HE-stained slice image, and the cancer histological subtype prediction result of the HE-stained slice image is used to indicate whether it belongs to a tumor with strong intratumoral heterogeneity (such as hepatocellular carcinoma) histological subtype.

[0041] S5, for each HE-stained slice image in the sample set, based on the comprehensive loss value of the attention network, the hybrid density network and the classifier, the hyperparameters are adjusted to generate a trained cancer histological subtype classification model.

[0042] In step S5, the training process realizes end-to-end training by optimizing the joint loss function, which includes three parts: classification loss, attention regression loss and MDN likelihood loss. Among them, the binary cross-entropy is used to calculate the final subtype prediction result. The attention regression loss minimizes the difference between the regression prediction value and the target value through the mean square error (MSE). The negative log-likelihood (NLL) loss of MDN is used to learn the mixing parameters. The total loss is the weighted combination of the three parts. Specifically, the loss value can be calculated by the following formula: ; ; ; ; wherein, is the classification loss, is the attention regression loss, is the hybrid density loss, is the actual result of the cancer histological subtype, is the predicted result of the cancer histological subtype. 、 、 is the hyperparameter adjusted by the validation set.

[0043] The attention regression-enhanced multi-instance learning method (Regression-Enhanced Multi-Instance Learning, RE-MIL) for cancer histological subtype classification diagnosis based on HE images provided in the present application combines the multi-instance learning framework through the attention regression-enhanced mechanism to classify cancer pathological images, which can effectively extract key features from HE images, improve the generalization ability and accuracy of the classification model, and the subtype classification precision.

[0044] In an implementable embodiment, the HE-stained slice image to be classified can be input into the pre-trained cancer histological subtype classification model to obtain a cancer histological subtype classification result output by the cancer histological subtype classification model, and the cancer histological subtype classification result is used to indicate whether the HE-stained slice image to be classified belongs to the target cancer histological subtype. In this way, the predicted cancer histological subtype classification result is more accurate.

[0045] Based on the same inventive concept, the application further provides a training device for a cancer histological subtype classification model, which comprises a training module and is specifically configured to: for each HE-stained slice image in a sample set, identify a tumor image region in the HE-stained slice image and divide a plurality of non-overlapping sub-slice images in the tumor image region; for each sub-slice image corresponding to each HE-stained slice image in the sample set, input the sub-slice image into a feature extraction network to obtain a basic feature vector of the sub-slice image; for each HE-stained slice image in the sample set, input the basic feature vectors of all sub-slice images of the HE-stained slice image into a connected attention network and a hybrid density network to construct an enhanced feature vector of the HE-stained slice image; for each HE-stained slice image in the sample set, input the enhanced feature vector of the HE-stained slice image into a classifier to output a cancer histological subtype prediction result of the HE-stained slice image; and for each HE-stained slice image in the sample set, adjust hyperparameters based on a comprehensive loss value of the attention network, the hybrid density network and the classifier to generate a trained cancer histological subtype classification model.

[0046] In a preferred embodiment, the training module constructs the enhanced feature vector of the target HE-stained slice image in the following manner: The training module inputs the basic feature vectors of all sub-slice images of the target HE-stained slice image into the attention network to obtain an attention score and an optimized feature vector of each sub-slice image output by the attention network. The training module inputs the attention score and the optimized feature vector of each sub-slice image of the target HE-stained slice image into the hybrid density network to obtain the enhanced feature vector of the target HE-stained slice image output by the hybrid density network.

[0047] In a preferred embodiment, the training module is specifically configured to input the basic feature vectors of all sub-slice images of the target HE-stained slice image into the attention network to calculate the attention score of each sub-slice image. Based on the attention score of each sub-slice image of the target HE-stained slice image, the training module determines a weight coefficient value corresponding to each sub-slice image. For each sub-slice of the target HE-stained section image, the optimized feature vector of the sub-slice image is calculated based on the basic feature vector of the sub-slice image and the corresponding weight coefficient value.

[0048] In a preferred embodiment, the training module is specifically used to input the attention scores and optimized feature vectors of all sub-slice images of the target HE staining slice image into the mixing density network to calculate the mixing coefficient values, mean and standard deviation values ​​of all sub-slice images. For each sub-slice image of the target HE staining section image, the cluster assignment probability value of the sub-slice image is calculated based on the attention score, mixing coefficient value, mean and standard deviation value of the sub-slice image; Based on the class assignment probability value and mean of each sub-slice image, the regression prediction value of that sub-slice image is calculated; The enhanced feature vector is obtained by weighting the optimized feature vectors and regression predictions of all sub-slice images of the target HE-stained slice image.

[0049] In a preferred embodiment, the training module calculates the cluster assignment probability value using the following formula: ; in, For the first The sub-slice image belongs to the first The probability values ​​assigned to each cluster. The first output of the attention network Attention scores for individual slice images, The output of the hybrid density network The mean coefficients corresponding to each sub-slice image. The output of the hybrid density network The square of the standard deviation coefficients corresponding to each sub-slice image The output of the hybrid density network The mixing coefficients corresponding to each sub-slice image; Furthermore, the regression prediction value is calculated using the following formula: ; in, For the first The regression prediction value corresponding to each sub-slice image.

[0050] In a preferred embodiment, the training module calculates the loss value using the following formula: ; ; ; ; wherein, is a classification loss, is an attention regression loss, is a mixture density loss, is a cancer histological subtype actual result, is a cancer histological subtype predicted result.

[0051] In a preferred embodiment, the training module determines the weight coefficient value corresponding to the sub-slice image by the following formula: ; wherein, is an attention score corresponding to the i-th sub-slice image.

[0052] Referring to Figure 3 , Figure 3 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. As shown in the Figure 3 , the electronic device 300 includes a processor 310, a memory 320 and a bus 330.

[0053] The memory 320 stores machine readable instructions executable by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate through the bus 330. The machine readable instructions executed by the processor 310 can perform the steps of the training method of the classification model of the cancer histological subtype in the method embodiment as described above Figure 1 . For specific implementation, please refer to the method embodiment, which will not be repeated here.

[0054] The present application also provides a computer readable storage medium, which stores a computer program. When the computer program is run by a processor, the steps of the training method of the classification model of the cancer histological subtype in the method embodiment as described above Figure 1 . For specific implementation, please refer to the method embodiment, which will not be repeated here.

[0055] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

[0056] ​In the embodiments of the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. The embodiments described above are merely exemplary, for example, the division of the units is only a logical function division, and there can be another division manner in actual implementation; for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0057] In addition, the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0058] In addition, the functional modules in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0059] It should be noted that if the functions are realized in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0060] In this document, the terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations.

[0061] The above merely provides an example of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for training a classification model of cancer histological subtypes, characterized in that, The method includes: For each HE-stained slide image in the sample set, the tumor image region in the HE-stained slide image is identified, and multiple non-overlapping sub-slice images are divided within the tumor image region. For each sub-slice image corresponding to each HE-stained slice image in the sample set, the sub-slice image is input into the feature extraction network to obtain the basic feature vector of the sub-slice image. For each HE-stained slice image in the sample set, the basic feature vectors of all sub-slice images of the HE-stained slice image are input into an interconnected attention network and a hybrid density network to construct the enhanced feature vector of the HE-stained slice image. For each HE-stained slide image in the sample set, the enhanced feature vector of the HE-stained slide image is input into the classifier to output the cancer histological subtype prediction result of the HE-stained slide image. For each HE-stained slide image in the sample set, the hyperparameters are adjusted based on the combined loss values ​​of the attention network, the mixed density network, and the classifier to generate a trained cancer histological subtype classification model.

2. The method according to claim 1, characterized in that, The enhanced feature vector of the target HE-stained slice image is constructed using the following method: The basic feature vectors of all sub-slice images of the target HE-stained slice image are input into the attention network to obtain the attention score and optimized feature vector of each sub-slice image output by the attention network; The attention scores and optimized feature vectors of all sub-slices of the target HE-stained slice image are input into a hybrid density network to obtain the enhanced feature vector of the target HE-stained slice image output by the hybrid density network.

3. The method according to claim 2, characterized in that, The steps for optimizing the feature vector of each sub-slice image output by the attention network specifically include: The basic feature vectors of all sub-slice images of the target HE-stained slice image are input into the attention network to calculate the attention score of each sub-slice image; Based on the attention scores of all sub-slice images of the target HE staining slice image, the weight coefficient value corresponding to each sub-slice image is determined; For each sub-slice of the target HE-stained section image, the optimized feature vector of the sub-slice image is calculated based on the basic feature vector of the sub-slice image and the corresponding weight coefficient value.

4. The method according to claim 3, characterized in that, The steps of outputting the enhanced feature vector of the target HE-stained slice image using a hybrid density network specifically include: The attention scores and optimized feature vectors of all sub-slice images of the target HE staining slice image are input into the mixing density network to calculate the mixing coefficients, mean and standard deviation of all sub-slice images. For each sub-slice image of the target HE staining section image, the cluster assignment probability value of the sub-slice image is calculated based on the attention score, mixing coefficient value, mean and standard deviation value of the sub-slice image; Based on the class assignment probability value and mean of each sub-slice image, the regression prediction value of that sub-slice image is calculated; The enhanced feature vector is obtained by weighting the optimized feature vectors and regression predictions of all sub-slice images of the target HE-stained slice image.

5. The method according to claim 4, characterized in that, The cluster assignment probability value is calculated using the following formula: ； in, For the first The sub-slice image belongs to the first The probability values ​​assigned to each cluster. The first output of the attention network Attention scores for individual slice images, The output of the hybrid density network The mean coefficients corresponding to each sub-slice image. The output of the hybrid density network The square of the standard deviation coefficients corresponding to each sub-slice image The output of the hybrid density network The mixing coefficients corresponding to each sub-slice image; Furthermore, the regression prediction value is calculated using the following formula: ； in, For the first The regression prediction value corresponding to each sub-slice image.

6. The method according to claim 1, characterized in that, The loss value is calculated using the following formula: ； ； ； ； in, For classifying losses, For attention regression loss, For the mixed density loss, For the actual results of cancer histological subtypes, The results are for predicting cancer histological subtypes.

7. The method according to claim 3, characterized in that, The weight coefficient values ​​corresponding to the sub-slice images are determined using the following formula: ； in, For the first Attention scores corresponding to each sub-slice image.

8. The method according to claim 1, characterized in that, The HE-stained slide image to be classified is input into a pre-trained cancer histology subtype classification model to obtain the cancer histology subtype classification result output by the cancer histology subtype classification model. The cancer histology subtype classification result is used to indicate whether the HE-stained slide image to be classified belongs to the target cancer histology subtype.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is in operation, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the training method for a classification model of a cancer histological subtype as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of a training method for a classification model of any of the cancer histological subtypes as described in any one of claims 1 to 8.

Images