Histopathological Identification Method and Device for Lymph Node Metastasis of Prostate Cancer

Abstract

The present application discloses a histopathological recognition method and device for prostate cancer lymph node metastasis. The method includes: performing pathological recognition on the pathological image of the prostate lymph node to obtain a first confidence level corresponding to the original pathological recognition result; performing masking processing on the pathological image to obtain an original pathological mask image; correcting the first confidence level according to a plurality of pathological example images in a preset pathological image library to obtain a target confidence level; when the original pathological mask image includes a target tissue mask part, performing pixel correction on the target tissue mask part according to the target confidence level, and outputting a target pathological mask image. By correcting the first confidence level according to a plurality of pathological example images in the preset pathological image library, the confidence level error caused by insufficient generalization performance in pathological recognition can be reduced, thereby improving the accuracy of the output target pathological mask image.

Description

Technical Field

[0001] This application relates to the field of medical technology, and in particular, to a histopathological identification method and device for lymph node metastasis of prostate cancer. Background Art

[0002] Lymph node metastasis often occurs in prostate cancer. Currently, when treating prostate cancer patients, radical prostatectomy is often performed simultaneously with pelvic lymph node dissection, and histopathological section examination of the excised lymph nodes is carried out to observe whether there are cancer cells in the lymph node tissue to determine whether prostate cancer has metastasized to the lymph nodes, so as to decide the next treatment strategy. However, the pathological review of lymph nodes is a monotonous and heavy visual recognition task, and the human eye is prone to fatigue and is likely to miss small-diameter lymph node metastasis foci. The relative shortage of pathologists nationwide and the heavy workload of pathologists have exacerbated this phenomenon. Previous studies have shown that the missed diagnosis rate of conventional pathological reports for micro-metastasis foci (<2 mm) of prostate cancer lymph nodes is as high as 8-13%. This greatly affects the diagnosis and treatment quality of prostate cancer patients.

[0003] In order to improve the accuracy of pathological diagnosis of prostate cancer lymph node metastasis and reduce the missed diagnosis rate of prostate cancer lymphatic micro-metastasis foci, image recognition technology can be introduced to identify prostate lymph node pathological tissues to assist pathologists in pathological observation. However, in the existing technology, there are few identification technologies for lymph node pathological tissues, and the existing identification technologies for lymph node pathological tissues are relatively simple and the accuracy is average. When using the existing identification technology for lymph node pathological tissues to identify prostate cancer lymph node metastasis foci, missed diagnosis is likely to occur. Summary of the Invention

[0004] The following is an overview of the subject matter described in detail in this article. This overview is not intended to limit the scope of protection of the claims.

[0005] Embodiments of this application provide a histopathological identification method and device for lymph node metastasis of prostate cancer, which can improve the identification accuracy of prostate lymph node pathological tissues.

[0006] On the one hand, embodiments of this application provide a histopathological identification method for lymph node metastasis of prostate cancer, including the following steps:

[0007] Perform pathological identification based on the pathological image of the prostate lymph node to obtain the first confidence corresponding to the original pathological identification result;

[0008] Perform masking processing on the pathological image to obtain the original pathological mask image;

[0009] Correct the first confidence according to multiple pathological example images in a preset pathological image library to obtain the target confidence;

[0010] When the original pathological mask image includes a target tissue mask part, pixel correction is performed on the target tissue mask part according to the target confidence level to output a target pathological mask image.

[0011] Optionally, multiple of the pathological example images form multiple pathological image classification clusters in the preset pathological image library;

[0012] The obtaining of the first confidence level corresponding to the original pathological recognition result by performing pathological recognition on the pathological image includes:

[0013] Performing feature extraction on the pathological image to obtain a transitional pathological feature image;

[0014] The obtaining of the target confidence level by correcting the first confidence level according to multiple pathological example images in the preset pathological image library includes:

[0015] Performing cluster classification on the transitional pathological feature image according to multiple of the pathological image classification clusters, and determining a target pathological image classification cluster from multiple of the pathological image classification clusters, where the target pathological image classification cluster includes multiple of the pathological example images;

[0016] Correcting the first confidence level according to multiple of the pathological example images in the target pathological image classification cluster to obtain the target confidence level.

[0017] Optionally, the preset pathological image library includes a pathological image classification label corresponding to each of the pathological image classification clusters, and a second confidence level corresponding to each of the pathological example images, where the second confidence level is a pathological tissue recognition confidence level obtained by performing pathological recognition on the pathological example image;

[0018] The obtaining of the target confidence level by correcting the first confidence level according to multiple of the pathological example images in the target pathological image classification cluster includes:

[0019] Determining a target preset correction offset value corresponding to the target pathological image classification cluster from multiple preset correction offset values according to the pathological image classification label of the target pathological image classification cluster, where multiple of the preset correction offset values correspond one-to-one to multiple of the pathological image classification clusters;

[0020] Correcting the first confidence level according to multiple of the second confidence levels corresponding to the target pathological image classification cluster and the target correction offset value to obtain the target confidence level.

[0021] Optionally, correcting the first confidence level according to the multiple second confidence levels corresponding to the target pathological image classification cluster and the target correction offset value to obtain the target confidence level includes:

[0022] Determining a third confidence level and a fourth confidence level among the multiple second confidence levels corresponding to the target pathological image classification cluster;

[0023] Correcting the first confidence level according to the third confidence level, the fourth confidence level and the target correction offset value to obtain the target confidence level.

[0024] Optionally, pixel-correcting the target tissue mask part according to the target confidence level and outputting a target pathological mask image includes:

[0025] Determining a target pathological recognition result according to the target confidence level;

[0026] When the target pathological recognition result indicates that the prostate lymph node includes the target tissue, generating a plurality of target pixels according to the target confidence level and the target tissue mask part;

[0027] Performing pixel replacement on the target tissue mask part according to the plurality of target pixels and outputting the target pathological mask image.

[0028] Optionally, the target tissue mask part includes a plurality of original pixels;

[0029] The generating a plurality of target pixels according to the target confidence level and the target tissue mask part includes:

[0030] For each of the original pixels, adjusting the pixel value of the original pixel according to the target confidence level to obtain a target pixel value corresponding to the original pixel;

[0031] Generating a plurality of the target pixels according to the plurality of target pixel values.

[0032] Optionally, the pathological image is obtained by acquiring a pathological section image of a prostate lymph node and performing a sliding window cropping process on the pathological section image, wherein a plurality of pathological images are obtained by performing the sliding window cropping process on the pathological section image;

[0033] The method further includes:

[0034] Performing pathological recognition on the next pathological image to output the target pathological mask image corresponding to the next pathological image;

[0035] After obtaining the target pathological mask images corresponding to the multiple pathological images, perform graphic merging based on the target pathological mask images corresponding to the multiple pathological images to obtain the target pathological section mask image corresponding to the pathological section image, where the target pathological section mask image includes the target tissue body mask of the prostate lymph nodes.

[0036] On the other hand, an embodiment of the present application further provides a histopathological recognition device for prostate cancer lymph node metastasis, including:

[0037] An original pathological recognition module, configured to perform pathological recognition based on the pathological image of the prostate lymph nodes to obtain the first confidence level corresponding to the original pathological recognition result;

[0038] An original pathological mask processing module, configured to perform mask processing on the pathological image to obtain an original pathological mask image;

[0039] A target confidence level acquisition module, configured to correct the first confidence level according to a plurality of pathological example images in a preset pathological image library to obtain a target confidence level;

[0040] A target pathological mask image output module, configured to, when the original pathological mask image includes a target tissue mask part, perform pixel correction on the target tissue mask part according to the target confidence level and output a target pathological mask image.

[0041] On the other hand, an embodiment of the present application further provides an electronic device, including:

[0042] At least one processor;

[0043] At least one memory, configured to store at least one program;

[0044] When at least one of the programs is executed by at least one of the processors, the method described above is implemented.

[0045] On the other hand, an embodiment of the present application further provides a computer-readable storage medium, in which a computer program executable by a processor is stored, and when the computer program executable by the processor is executed by the processor, it is used to implement the method described above.

[0046] On the other hand, an embodiment of the present application further provides a computer program product, including a computer program or computer instructions, the computer program or the computer instructions are stored in a computer-readable storage medium, a processor of an electronic device reads the computer program or the computer instructions from the computer-readable storage medium, and the processor executes the computer program or the computer instructions, so that the electronic device executes the method described above.

[0047] The embodiments of the present application at least include the following beneficial effects: First, pathological recognition can be performed on the pathological images of prostate lymph nodes to obtain the first confidence level corresponding to the original pathological recognition result. Then, the pathological images are subjected to masking processing to obtain the original pathological mask images. Next, the first confidence level is corrected according to multiple pathological example images in the preset pathological image library to obtain the target confidence level. When the original pathological mask image includes the target tissue mask part, pixel correction is performed on the target tissue mask part according to the target confidence level, and the target pathological mask image is output. By correcting the first confidence level with multiple pathological example images in the preset pathological image library, the error in the confidence level corresponding to the lymph node metastasis foci that are difficult to recognize due to insufficient generalization performance of pathological recognition can be reduced, so that the target confidence level is closer to the true recognition result, enabling the difficult-to-recognize lymph node metastasis foci to be recognized more accurately, and it can be used as a reliable reference for the original pathological mask image. Then, pixel correction is performed on the target tissue mask part through the target confidence level, so that the target pathological mask image can correspond to the target confidence level, thereby enabling the target pathological mask image to more accurately display the difficult-to-recognize lymph node metastasis foci, and further making the target pathological mask image closer to the true recognition result, providing a more accurate recognition result for medical staff, and reducing the situation where medical staff need to spend extra time to determine whether the recognition result of the pathological tissue of prostate lymph nodes is correct.

[0048] Other features and advantages of the present application will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present application. The objectives and other advantages of the present application can be achieved and obtained through the structures specifically pointed out in the specification and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The drawings are used to provide a further understanding of the technical solutions of the present application, and constitute a part of the specification. They are used together with the embodiments of the present application to explain the technical solutions of the present application, and do not constitute a limitation to the technical solutions of the present application.

[0050] Figure 1 is a flowchart of a method for histopathological recognition of prostate lymph node metastasis provided by an embodiment of the present application;

[0051] Figure 2 is an execution architecture of a method for histopathological recognition of prostate lymph node metastasis provided by a specific embodiment of the present application;

[0052] Figure 3 is an execution architecture of a method for histopathological recognition of prostate lymph node metastasis provided by another specific embodiment of the present application;

[0053] Figure 4 is a schematic diagram of enhancing the target tissue mask part provided by an embodiment of the present application;

[0054] Figure 5 is a schematic diagram of deleting a target tissue mask portion provided by an embodiment of the present application;

[0055] Figure 6 yes Figure 1 A flowchart of sub-steps of a specific embodiment of step 103;

[0056] Figure 7 yes Figure 6 A flowchart of sub-steps of a specific embodiment of step 602;

[0057] Figure 8 yes Figure 7 A flowchart of sub-steps of a specific embodiment of step 702;

[0058] Figure 9 yes Figure 1 A flowchart of sub-steps of a specific embodiment of step 104;

[0059] Figure 10 yes Figure 9 A flowchart of sub-steps of a specific embodiment of step 902;

[0060] Figure 11 is a schematic diagram of a tissue pathology identification device provided in an embodiment of the present application;

[0061] Figure 12 This is a schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The present application is further described below in conjunction with the accompanying drawings and specific embodiments. The described embodiments should not be considered as limiting the present application. All other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0063] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0065] In order to improve the accuracy of pathological diagnosis of lymph node metastasis in prostate cancer and reduce the missed diagnosis rate of lymphatic micrometastasis lesions in prostate cancer, image recognition technology can be introduced to identify the pathological tissues of prostate lymph nodes to assist pathologists in pathological observation. However, in the existing technology, there are few identification technologies for lymph node pathological tissues. The existing identification technologies for lymph node pathological tissues are relatively simple and have average accuracy. When using the existing identification technologies for lymph node pathological tissues to perform lymph node pathological film reading and identification on prostate cancer lymph node metastases, missed diagnosis is likely to occur.

[0066] In order to improve the identification accuracy of prostate lymph node pathological tissues, the embodiments of the present application provide a histopathological identification method for prostate lymph node metastasis, a histopathological identification device for prostate lymph node metastasis, an electronic device, a computer-readable storage medium, and a computer program product. The histopathological identification method can first perform pathological identification based on the pathological image of the prostate lymph node to obtain the first confidence corresponding to the original pathological identification result. Then, perform masking processing on the pathological image to obtain the original pathological mask image. Next, correct the first confidence according to multiple pathological example images in the preset pathological image library to obtain the target confidence. When the original pathological mask image includes the target tissue mask part, perform pixel correction on the target tissue mask part according to the target confidence to output the target pathological mask image. By correcting the first confidence according to multiple pathological example images in the preset pathological image library, the error in the confidence corresponding to the lymph node metastases that are difficult to identify due to insufficient generalization performance of pathological identification can be reduced, so that the target confidence is closer to the true identification result, making the lymph node metastases that are difficult to identify more accurately identified, and can be used as a reliable reference for the original pathological mask image. Then, perform pixel correction on the target tissue mask part through the target confidence, so that the target pathological mask image can correspond to the target confidence, thereby enabling the target pathological mask image to more accurately display the lymph node metastases that are difficult to identify, and further making the target pathological mask image closer to the true identification result, providing a more accurate identification result for medical staff and reducing the situation where medical staff need to spend extra time to determine whether the identification result of prostate lymph node pathological tissues is correct.

[0067] Referring to Figure 1 shown, Figure 1The figure is a flowchart of a histopathological recognition method for prostate lymph node metastasis provided by an embodiment of the present application. This histopathological recognition method can be executed by a server, or by an electrical observation device such as an electron microscope, or can be jointly executed by an electrical observation device and a server. In the embodiment of the present application, taking the method being executed by the server as an example for illustration, other execution examples can refer to the following embodiments. In the embodiment of the present application, the histopathological recognition method for prostate lymph node metastasis includes but is not limited to the following steps.

[0068] Step 101: Perform pathological recognition based on the pathological image of the prostate lymph node to obtain the first confidence level corresponding to the original pathological recognition result;

[0069] Step 102: Perform masking processing on the pathological image to obtain the original pathological mask image;

[0070] Step 103: Correct the first confidence level according to multiple pathological example images in the preset pathological image library to obtain the target confidence level;

[0071] Step 104: When the original pathological mask image includes the target tissue mask part, perform pixel correction on the target tissue mask part according to the target confidence level, and output the target pathological mask image.

[0072] In one embodiment, the pathological image of the prostate lymph node is obtained by a certain acquisition method. Among them, the acquisition method can be imaging the pathological section through an electrical observation device, or receiving the pathological image sent by the electrical observation device, or calling the pathological image of the prostate lymph node to be processed uploaded and stored in the database, etc., which is not specifically limited here. In addition, the pathological image of the prostate lymph node can be a pathological section image of the prostate lymph node, or a part of the pathological section image, which is not specifically limited here.

[0073] In addition, since the pathological image of the prostate lymph nodes is part of the pathological section image, the pathological image can be obtained from the overall pathological section image in the following ways, or can be obtained from the pathological section image by other methods, which are not specifically limited here. For example, one or more regions of interest are identified on the pathological section image by means of region of interest (ROI) identification, and the identified regions of interest are used as targets for image cutting, and the cut image is the pathological image. For another example, the target region is determined by obtaining external bounding information. Specifically, when medical staff are interested in a certain part of the pathological section image, the region of interest can be selected by bounding, and the device will form external bounding information according to the selected region, and the pathological section image is cut according to the external bounding, and the obtained image is the pathological image. For another example, the entire pathological section image is cut according to a preset cutting size, and all the cut images are pathological images.

[0074] In addition, the obtained pathological image can be the original image without image preprocessing, or the original pathological image after image preprocessing, which is not specifically limited here. Among them, the image preprocessing can be image filtering, image smoothing, grayscale conversion, etc., which are not specifically limited here.

[0075] In addition, the original pathological recognition result refers to the recognition result finally output in step 101 for pathological recognition. This recognition result can represent one or more target tissues on the pathological image. For example, if there is only one target tissue in the prostate lymph nodes on the pathological image, the original pathological recognition result can represent this target tissue; for another example, if there are multiple target tissues in the prostate lymph nodes on the pathological image, the original pathological recognition result can represent multiple target tissues. In the process of pathological recognition, multi-classification is performed to obtain multiple classification confidence levels, and the finally output recognition result is the result corresponding to the maximum confidence level among the multiple classification confidence levels (i.e., the original pathological recognition result), and the confidence level corresponding to this finally output recognition result is the first confidence level corresponding to the original pathological recognition result. Among them, the original pathological recognition result can be a binary classification result, a ternary classification result, a quaternary classification result, etc., which are not specifically limited here. In addition, the number of the first confidence levels corresponds to the number of target tissues represented by the original pathological recognition result, that is, each recognized target tissue corresponds to a first confidence level. For the convenience of subsequent embodiment description, the following embodiments are described by taking the original pathological recognition result as a binary classification result, which is used to indicate whether the prostate lymph nodes include target tissues, there is only one target tissue in the prostate lymph nodes on the pathological image, and there is only one first confidence level as an example. Other situations such as multiple target tissues in the prostate lymph nodes and multiple first confidence levels can be executed one by one with reference to the following embodiments, and no more details are given here.

[0076] In one embodiment, the target tissue can be pathological tissue on the prostate lymph nodes or healthy tissue on the prostate lymph nodes, and no specific limitation is made here. For the convenience of the description of subsequent embodiments, the subsequent embodiments will further take the target tissue as pathological tissue as an example for description (that is, the original pathological recognition result is used to indicate whether there is pathological tissue in the prostate lymph nodes). The embodiments with the target tissue as healthy tissue can be replaced with reference to the following embodiments and will not be elaborated hereinafter.

[0077] In one embodiment, pathological recognition can be achieved in the following manner: First, input the pathological image of the prostate lymph nodes into the first network. The first network can first perform feature extraction on the pathological image of the prostate lymph nodes to obtain the first feature image of the pathological image, and then perform feature processing such as pooling and depth convolution on the first feature image to normalize the information in the first feature image, obtain low-dimensional first feature information, calculate the classification confidence through an activation function to obtain multiple classification confidences, and use the recognition result corresponding to the highest classification confidence value among the multiple classification confidences as the original pathological recognition result. Moreover, use the highest classification confidence as the first confidence. Among them, the first network can be a Convolutional Neural Network (CNN), a Convolutional Recurrent Neural Network (CRNN), etc., and no specific limitation is made here.

[0078] In one embodiment, mask processing can be achieved in the following manner: First, input the pathological image of the prostate lymph nodes into the second network. The second network can first perform feature extraction on the pathological image of the prostate lymph nodes to obtain the second feature image of the pathological image of the prostate lymph nodes, and then perform upsampling on the second feature image to perform pixel prediction. Then, classify the pixels in the feature image after upsampling through multiple convolutional layers, and generate mask information on the pathological image according to the pixel classification result, so as to obtain the original pathological mask image. Among them, the second network can be a U-Net, a DeepLab, etc., and no specific limitation is made here. In addition, when the mask processing can classify the pixels related to the target tissue, the original pathological mask image can carry the mask information of the target tissue and other mask information such as background mask information; when the mask processing cannot classify the pixels related to the target tissue, the original pathological mask image can carry other mask information such as background mask information, and no specific limitation is made here.

[0079] In one embodiment, the preset pathological image library can be an ordinary database, or a database that combines the functions of ordinary data storage and vector storage, which is not specifically limited here. In addition, the pathological example image can be a typical pathological image of the prostate lymph node, or a feature image obtained after feature extraction from the typical pathological image of the prostate lymph node, etc., which is not specifically limited here. Among them, the mentioned typical pathological image can be a pathological image such as a pathological image with a target tissue having a relatively large area ratio, a pathological image with a relatively concentrated target tissue, a pathological image with a target tissue having a relatively good staining effect, a pathological image in which there are obvious morphological differences between the cells in the target tissue and normal cells, a pathological image in which there are obvious morphological differences between the cell nuclei of the cells in the target tissue and the cell nuclei of normal cells, a pathological image in which the cells in the target tissue are arranged disorderly, etc., which belong to pathological images of easy samples, or a pathological image such as a pathological image with a target tissue having a relatively small area ratio, a pathological image with a relatively dispersed target tissue, a pathological image with a target tissue having a relatively poor staining effect, a pathological image in which there are relatively small morphological differences between the cells in the target tissue and normal cells under specific circumstances, a pathological image in which there are relatively small morphological differences between the cell nuclei of the cells in the target tissue and the cell nuclei of normal cells under specific circumstances, a pathological image in which there are relatively small differences in the arrangement structure between the cell arrangement structure of the target tissue and that of the normal tissue under specific circumstances, etc., which belong to pathological images of hard samples, which is not specifically limited here.

[0080] Referring to Figure 2 as shown, Figure 2 illustrates the execution architecture of a method for histopathological identification of prostate lymph node metastasis in a specific embodiment. In one embodiment, parallel networks are used in steps 101 and 102, and different-precision feature extraction methods are used for pathological identification and mask processing respectively. Referring to the implementation methods of the above-mentioned pathological identification and mask processing, step 101 uses the first feature image to complete the pathological identification process and obtains the first confidence level, and step 102 uses the second feature image to complete the mask processing process and obtains the original pathological mask graph. Different image processing tasks have different requirements for the accuracy of the required image information. If the same feature image is used, for a certain execution network, it may cause the loss of image information, thereby affecting the execution accuracy of at least one of pathological identification and mask processing. By using an execution architecture with parallel networks, both pathological identification and mask processing can obtain the image information (i.e., feature images) with the required accuracy for each of them, so that the execution networks of pathological identification and mask processing can respectively output a relatively high-reliability first confidence level and a relatively high-accuracy original pathological mask image. Furthermore, a more accurate target pathological mask image can be output subsequently, thereby improving the pathological judgment efficiency of medical staff for prostate lymph nodes.

[0081] Referring to Figure 3 as shown Figure 3 shown is the execution architecture of the histopathological identification method for prostate lymph node metastasis in another specific embodiment. In one embodiment, the network parts used in step 101 and step 102 are combined. Specifically, in Figure 3 the shown execution architecture, the pathological image can be first subjected to feature extraction to obtain a transitional pathological feature image. For the masking process, upsampling can be performed according to the transitional pathological feature image, and pixel prediction can be performed through upsampling to increase the receptive field. Then, each pixel in the transitional pathological feature image after upsampling processing can be classified through multiple convolutional layers, and a mask can be formed on the pathological image according to the classification of each pixel to obtain the original pathological mask image; for pathological identification, the number of features of the transitional pathological feature image can be unified to obtain a third feature image, and then the healthy tissue and pathological tissue of the prostate lymph nodes on the pathological image can be classified according to the third feature image, and the classification confidence of each classification can be obtained. The maximum classification confidence is determined as the first confidence corresponding to the original pathological identification result. Generally speaking, compared with pathological identification, the masking process has a lower accuracy requirement for image information. Therefore, the transitional feature image can be first extracted based on the accuracy requirement of the feature image of the masking process. At this time, the transitional feature image meets the accuracy of steps such as upsampling in the masking process to output the original pathological mask image; and because the accuracy requirement of the feature image for pathological identification is higher, feature extraction can be performed on the basis of the transitional pathological feature image to obtain a third pathological feature image with higher accuracy to meet the accuracy requirement of the image information required for pathological identification. By adopting Figure 3 the combined execution structure shown in

[0082] For the convenience of the following description, the following embodiments will be described based on Figure 3 the execution architecture in Figure 2 Embodiments based on the execution structure in

[0083] can refer to the following embodiments and will not be elaborated here.In one embodiment, the pathological example image is a historical transitional pathological feature image after steps 101 to 104. Each pathological example image stores a corresponding first evaluation score and second evaluation score in the preset pathological image library. Among them, the first evaluation score can be used to evaluate the gap between the target pathological mask image corresponding to the pathological example image and the original pathological mask image, and the second evaluation score can be used to evaluate the gap between the target pathological mask image of the pathological example image and the true pathological mask image. The true pathological mask image is an artificial pathological mask image corrected by medical staff based on the target pathological mask image. If the medical staff does not correct the target pathological mask image, the second evaluation score is 0. Specifically, after each pathological image goes through steps 101 to 104, the first evaluation score and the second evaluation score corresponding to the pathological image are calculated according to the transitional pathological feature image, and then the transitional pathological feature image is stored in the preset pathological image library as a pathological example image, and its first evaluation score and second evaluation score are stored in the preset pathological image library correspondingly; during the generation of the target pathological mask image corresponding to the new pathological image, multiple pathological example images are randomly selected from the preset pathological image library, and the corresponding first evaluation score and second evaluation score of these pathological example images are determined, and the first confidence level of the new pathological image is corrected based on these first evaluation scores and second evaluation scores to obtain the target confidence level.

[0084] In one embodiment, in step 103, one or more similar pathological example images are determined from multiple pathological example images according to the transitional feature image, and the first confidence level is decreased or increased according to whether these pathological example images represent that the prostate lymph node tissue includes the target tissue, so as to obtain the target confidence level.

[0085] In the pathological section of the prostate lymph nodes, due to the fact that the target tissue may have various forms, be scattered in position, have poor staining effects, and there are significant differences between the color distribution domain of the dataset used during the training of the execution architecture and the color distribution domain of the pathological images during actual pathological reading, the original pathological recognition result may not be accurate (i.e., the first confidence level is unreliable), resulting in the non-correspondence between the original pathological mask image and the original pathological recognition result, thus casting doubt on the accuracy of the original pathological mask image. For example, the target tissue occupies a relatively small area in the prostate lymph node tissue, and the differentiation between the target tissue and the surrounding healthy tissue in the pathological image is not high. This information may be ignored during the pathological recognition process due to reasons such as deep convolution, and ultimately an incorrect original pathological recognition result may be obtained. Since upsampling is performed on the transitional feature processing image during the mask processing, the differentiation information between the target tissue and the surrounding healthy tissue can be enhanced. As a result, the original mask image can accurately show the target tissue included in the pathological image through the mask part of the target tissue. These situations will make the original mask image conflict with the previously obtained original pathological recognition result, instead reducing the credibility of the original pathological mask image. Another example is that due to uneven staining, different imaging parameters used by medical staff during the imaging operation of the section, etc., the imaging effects of the tissues on the pathological image are different, resulting in a large difference between the color distribution domain of the pathological image and the color distribution domain of the dataset used during the training of the execution architecture, ultimately leading to incorrect original pathological results and incorrect original pathological mask images, reducing the reliability of pathological recognition.

[0086] Therefore, the first confidence level can be corrected through similar pathological example images to improve the accuracy of the original pathological recognition result and obtain a more accurate pathological recognition result (i.e., the target pathological recognition result mentioned below). Relatively speaking, the reliability of the corrected target confidence level is also higher than the first confidence level. Therefore, by correcting the first confidence level through multiple pathological example images in the preset pathological image library, the accuracy of the original pathological recognition result can be corrected, reducing the recognition error caused by the error between the training set and the actual image in the execution architecture, thereby reducing the impact brought by the insufficient generalization performance of pathological recognition, and providing a reference for the accuracy of the original pathological mask image, enabling the original masked pathological image to be adjusted accordingly through the target confidence level, so that the difficult-to-recognize target tissue on the pathological image can be correctly displayed in the target pathological mask image, and thus enabling the final output target pathological mask image to be more accurate.

[0087] In one embodiment, referring to Figure 4As shown, in step 104, when the original pathology mask image includes a target tissue mask portion, and the target confidence indicates that the prostate lymph node includes the target tissue, pixel correction is performed on the target tissue mask portion to enhance the target tissue mask information in the original pathology mask image, so that the first confidence corresponding to the target tissue in the output target pathology mask image can be improved, thereby allowing medical staff to more easily and clearly see the target tissue in the prostate lymph node in the pathology image.

[0088] In one embodiment, referring to Figure 5 As shown, in step 104, when the original pathology mask image includes the target tissue mask portion, and the target confidence indicates that the prostate lymph node does not include the target tissue, the pixel values corresponding to the target tissue mask information in the original pathology mask image are reduced by performing pixel correction on the target tissue mask portion, so as to correct the prediction results of the portions of the original pathology mask image that were previously predicted to be the target tissue, so that these portions are not predicted to be the target tissue. This allows the output target pathology mask image to not carry erroneous target tissue mask information, thereby avoiding, to a certain extent, the need for medical staff to spend extra time manually judging the target tissue mask portion.

[0089] In one embodiment, when the original pathology mask image does not include the target tissue mask portion and the target confidence indicates that the prostate lymph nodes do not include the target tissue, the original pathology mask image is output as the target pathology mask image.

[0090] In one embodiment, when the original pathology mask image does not include the target tissue mask portion, but the target confidence indicates that the prostate lymph nodes contain the target tissue, the current pathology image can be cropped into multiple pathology sub-images according to a preset image size using image cropping methods such as sliding window cropping. Masking processing is performed on each pathology sub-image to obtain an original pathology mask sub-image corresponding to each pathology sub-image. Pixel correction is then performed on each original pathology mask sub-image based on the target confidence, and a target pathology mask sub-image corresponding to each pathology sub-image is output. The multiple target pathology mask sub-images are then merged to obtain a target pathology mask image. The merging process may include image synthesis, image filtering, and image edge smoothing, etc., which are not specifically limited herein.

[0091] In one embodiment, the preset pathological image library may store the correct recognition results corresponding to each pathological example image. Step 103 may be implemented in the following manner: calculate the similarity between the pathological image and each pathological example image respectively, rank each pathological example image according to the similarity in descending order, and determine the top K pathological example images as the target pathological example images, where K is a natural odd number greater than 0. Determine the proportion of the recognition results corresponding to the K target pathological example images, and determine the specific recognition result corresponding to the one with the larger proportion of the recognition results as the correction label. Correct the first confidence level with the highest similarity corresponding to the correction label and the correction label itself to obtain the target confidence level.

[0092] Referring Figure 6 as shown Figure 6 shows Figure 1 sub-steps of a specific embodiment of step 103 in the figure. In one embodiment, multiple pathological example images form multiple pathological image classification clusters in the preset pathological image library. Step 103 may include but is not limited to the following steps.

[0093] Step 601: Cluster-classify the transitional pathological feature image according to the multiple pathological image classification clusters, and determine the target pathological image classification cluster from the multiple pathological image classification clusters;

[0094] Step 602: Correct the first confidence level according to the multiple pathological example images in the target pathological image classification cluster to obtain the target confidence level.

[0095] In one embodiment, multiple pathological example images may form multiple pathological image classification clusters in the preset pathological image library in the following manner, or may form multiple pathological image classification clusters in other ways, which are not specifically limited here. For example, calculate the similarity between each pathological example image and other pathological example images, and classify each pathological example image according to the similarity between each pathological example image and other pathological example images to obtain multiple pathological image classification clusters. For another example, classify through the attributes of each pathological example image to obtain multiple pathological image classification clusters. Among them, the attribute may be whether the prostate lymph nodes in the pathological example image have the target tissue, or the prostate lymph nodes in the pathological example image have a specific type of target tissue, etc., which are not specifically limited here. In addition, the target pathological image classification cluster includes multiple pathological example images.

[0096] In one embodiment, the pathological image classification clusters are formed based on the similarity between multiple pathological example images. The cluster classification of the transitional pathological feature image can be achieved in the following manner: First, calculate the similarity between the transitional pathological feature image and each pathological example image, determine the pathological example image that is most similar to the transitional pathological feature image, and determine the pathological image classification cluster corresponding to the most similar pathological example image as the target pathological image classification cluster corresponding to the transitional pathological feature image.

[0097] In one embodiment, the pathological image classification clusters are formed by clustering multiple pathological example images based on a certain clustering algorithm. In this clustering method, the prostate lymph nodes in each pathological example image in each pathological image classification cluster may or may not include the target tissue, etc., and no specific limitation is made here. Among them, the clustering algorithm may be the K-means clustering algorithm, the Hierarchical Clustering algorithm, etc., and no specific limitation is made here. The cluster classification of the transitional pathological feature image can be achieved in the following manner: Use the same clustering algorithm to divide the transitional pathological feature image to classify it into one of the multiple pathological image classification clusters, and this pathological image classification cluster is the target pathological image classification cluster.

[0098] In one embodiment, the preset pathological image library may store the preset compensation value corresponding to each pathological example image and the corresponding correct recognition result, and the pathological image classification cluster is formed based on the similarity between multiple pathological example images. Step 602 may determine the target confidence level in the following manner: First, determine the pathological example image with the highest similarity among the multiple pathological example images in the target pathological image classification cluster. Then, determine whether the correct recognition result corresponding to this pathological example image corresponds to the original pathological recognition result. If the correct recognition result corresponding to this pathological example image indicates that the corresponding prostate lymph node does not include the target tissue, while the original pathological recognition result indicates that the corresponding prostate lymph node includes the target tissue, the first confidence level is punished and corrected according to the preset compensation value to obtain the target confidence level; if the correct recognition result corresponding to this pathological example image indicates that the corresponding prostate lymph node includes the target tissue, and the original pathological recognition result also indicates that the corresponding prostate lymph node includes the target tissue, the first confidence level is rewarded and corrected according to the preset compensation value to obtain the target confidence level; if the correct recognition result corresponding to this pathological example image indicates that the corresponding prostate lymph node includes the target tissue, while the original pathological recognition result indicates that the corresponding prostate lymph node does not include the target tissue, the first confidence level is punished and corrected according to the preset compensation value to obtain the target confidence level; if the correct recognition result corresponding to this pathological example image indicates that the corresponding prostate lymph node does not include the target tissue, and the original pathological recognition result also indicates that the corresponding prostate lymph node does not include the target tissue, the first confidence level is rewarded and corrected according to the preset compensation value to obtain the target confidence level.

[0099] In one embodiment, the preset pathological image library may store the similarity between each pathological example image and other pathological example images, and the pathological image classification clusters are formed based on the attribute of whether multiple pathological example images include the target tissue part. Step 602 may determine the target confidence in the following manner: First, determine the pathological example image with the highest similarity to the transitional pathological feature image and its specific similarity among the multiple pathological example images in the target pathological image classification cluster, and judge whether the attribute of the target pathological image classification cluster corresponds to the original pathological recognition result. If the attribute corresponding to the target pathological image classification cluster is that the pathological example image includes the target tissue part, and the original pathological recognition result indicates that the corresponding prostate lymph node includes the target tissue, increase the first confidence according to the similarity between the transitional pathological feature image and the most similar pathological example image to obtain the target confidence; if the attribute corresponding to the target pathological image classification cluster is that the pathological example image does not include the target tissue part, and the original pathological recognition result indicates that the corresponding prostate lymph node does not include the target tissue, increase the first confidence according to the similarity between the transitional pathological feature image and the most similar pathological example image to obtain the target confidence; if the attribute corresponding to the target pathological image classification cluster is that the pathological example image does not include the target tissue part, and the original pathological recognition result indicates that the corresponding prostate lymph node includes the target tissue, decrease the first confidence according to the similarity between the transitional pathological feature image and the most similar pathological example image to obtain the target confidence; if the attribute corresponding to the target pathological image classification cluster is that the pathological example image does not include the target tissue part, and the original pathological recognition result indicates that the corresponding prostate lymph node includes the target tissue, decrease the first confidence according to the similarity between the transitional pathological feature image and the most similar pathological example image to obtain the target confidence. Among them, the specific manner of increasing the first confidence is diverse. The similarity can be added to the first confidence by weighting to obtain the target confidence, or the equivalent value of the similarity can be determined through a certain equivalent formula and added to the first confidence to obtain the target confidence, etc., which is not specifically limited here. In addition, the manner of decreasing the first confidence is diverse. The similarity can be decreased to the first confidence by weighting to obtain the target confidence, or the equivalent value of the similarity can be determined through a certain equivalent formula and the equivalent value can be deducted from the first confidence, etc., which is not specifically limited here.

[0100] Refer to Figure 7 as shown in Figure 7 which shows Figure 6 the sub-steps of a specific embodiment of step 602. In one embodiment, the preset pathological image library may include the pathological image classification labels corresponding to each pathological image classification cluster, and the second confidence corresponding to each pathological example image. Step 602 may include the following steps.

[0101] Step 701: Determine the target preset correction offset value corresponding to the target pathological image classification cluster among multiple preset correction offset values according to the pathological image classification label of the target pathological image classification cluster;

[0102] Step 702: Correct the first confidence according to multiple second confidence levels corresponding to the target pathological image classification cluster and the target preset correction offset value to obtain the target confidence.

[0103] In one embodiment, the pathological image classification label can indicate whether the target tissue part is included in the pathological image classification cluster. The second confidence level is obtained by performing pathological recognition on the pathological example image and is the recognition confidence level corresponding to the pathological image classification label. For example, the pathological example image is recognized as not including the target tissue in the prostate lymph node during pathological recognition, and the recognition confidence level is 70%. However, the pathological image classification cluster of this pathological example image is that the prostate lymph node includes the target tissue, so the second confidence level corresponding to this pathological example image is 30%. In addition, multiple preset correction offset values correspond one-to-one with multiple pathological image classification clusters. The target preset correction offset value can be 0.1, 0.2, 0.3, 0.4, 0.5, etc. Those skilled in the art can determine the specific values of each preset correction offset value according to the actual situation. For example, determine the specific positive or negative situation of the target preset correction offset value according to the pathological image classification label of the target pathological image classification cluster. Here, it is not specifically limited. In addition, multiple preset correction offset values can be stored through a preset pathological image library or in other ways. Here, it is not specifically limited.

[0104] In order to enable pathological recognition to have a certain generalization performance, when training related networks or executing architectures, a certain recognition ability will be discarded. In this case, pathological recognition will actually have a certain bias towards certain data, with better judgment ability for certain types of data and worse judgment ability for certain types of data. For example, pathological recognition is more biased towards pathological images such as those with larger target tissues on the prostate lymph node and those with a color distribution domain similar to that of the training set. These pathological images are more likely to be accurately recognized, while pathological recognition has a lower bias towards pathological images such as those with smaller and scattered target tissues on the prostate lymph node and those with a large difference in color distribution domain from the training set. These pathological images are more difficult to be accurately recognized. This results in the actual instability of the recognition results of pathological recognition, and the error situations mentioned above will occur.

[0105] For the target pathological image classification cluster, since the target pathological image classification cluster includes various forms of pathological example images, the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition can be determined according to the second confidence of each pathological example image and the pathological image classification label corresponding to the target pathological image classification cluster. When the overall second confidence of the target pathological image classification cluster is lower, the recognition accuracy of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition is lower, and the original mask image and the original pathological recognition result are more likely to be incorrect. Therefore, the first confidence can be adjusted according to the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition, so that the influence brought by the compromise of the generalization performance of pathological recognition can be reduced through the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition, and a more reliable target confidence can be obtained.

[0106] Specifically, first, a preset correction offset value can be set for various pathological image classification clusters to represent the offset degree of the pathological images of the corresponding type of various pathological image classification clusters in the pathological recognition process. Then, according to the multiple second confidences corresponding to the target pathological image classification cluster, the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition is determined, and the first confidence is corrected according to the preset correction offset value and the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition to obtain the target confidence.

[0107] It should be noted that the recognition result corresponding to the obtained target confidence (i.e., the target pathological recognition result) can be contrary to or the same as the original pathological recognition result. If it is necessary to output the pathological recognition result, a more accurate target pathological recognition result can be output, which is not specifically limited here.

[0108] In an embodiment, step 702 can be implemented in the following manner: first, perform recognition difficulty analysis according to the multiple second confidences corresponding to the target pathological image classification cluster to obtain the comprehensive confidence corresponding to the target pathological image classification cluster, and then superimpose the comprehensive confidence and the target preset correction offset value onto the first confidence to correct the first confidence to obtain the target confidence. By quantifying the recognition difficulty through the comprehensive confidence, the first confidence can be more accurately corrected by the difficulty level of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition, so that the influence brought by the compromise of the generalization performance of pathological recognition can be reduced, and the recognition stability of the pathological images of the corresponding type of the target pathological image classification cluster in pathological recognition can be improved.

[0109] Refer to Figure 8 shown Figure 8 shows Figure 7Sub-steps of a specific embodiment of step 702, and step 702 may include the following steps.

[0110] Step 801: Determine a third confidence level and a fourth confidence level among the multiple second confidence levels corresponding to the target pathological image classification cluster;

[0111] Step 802: Correct the first confidence level according to the third confidence level, the fourth confidence level, and the target preset correction offset value to obtain the target confidence level.

[0112] In one embodiment, the third confidence level and the fourth confidence level may be two random confidence levels among the multiple second confidence levels corresponding to the target pathological image classification cluster, or may be two specific confidence levels among the multiple second confidence levels corresponding to the target pathological image classification cluster, which is not specifically limited here.

[0113] In one embodiment, the difference between the third confidence level and the fourth confidence level, and the target preset correction offset value can be superimposed on the first confidence level to obtain the target confidence level.

[0114] In one embodiment, step 402 can obtain the target confidence level through the following formula (1).

[0115] (1),

[0116] In formula (1), can represent the target confidence level;

[0117] can represent the first confidence level;

[0118] can represent the set of confidence levels formed by the multiple second confidence levels corresponding to the target classification cluster;

[0119] can represent the preset confidence level correction coefficient;

[0120] can represent the third confidence level;

[0121] can represent the fourth confidence level;

[0122] can represent the target preset correction offset value.

[0123] When the third confidence level and the fourth confidence level are respectively the minimum value and the maximum value among the multiple second confidence levels corresponding to the target pathological image classification cluster, the influence of the samples that are most difficult to correctly identify in the target pathological image classification cluster is introduced through the third confidence level into the correction process of the first confidence level, so that the correction process of the first confidence level can be carried out based on the gap between the transitional pathological feature image and the pathological example image corresponding to the third confidence level. Thus, the correction ability for the first confidence level can be improved, and furthermore, the influence on the first confidence level caused by the compromise of the pathological recognition generalization performance can be reduced, so that the reliability of the target confidence level can be improved.

[0124] Refer to Figure 9 as shown Figure 9 shows Figure 1 sub-steps of a specific embodiment of step 104 in

[0125] Step 901: Determine the target pathological recognition result according to the target confidence level;

[0126] Step 902: When the target pathological recognition result indicates that the prostate lymph node includes the target tissue, generate multiple target pixels according to the target confidence level and the target tissue mask part;

[0127] Step 903: Perform pixel replacement on the target tissue mask part according to the multiple target pixels, and output the target pathological mask image.

[0128] In one embodiment, on the premise that the original pathological mask image includes the target tissue mask part, when the target pathological recognition result indicates that the prostate lymph node includes the target tissue, it can be determined that the target tissue mask information in the original pathological mask image is correct. On this basis, multiple target pixels can be generated according to the target confidence level and the target tissue mask part, and pixel replacement is performed on the target tissue mask part, so as to improve the distinction between the target tissue mask part and other mask parts, so that medical staff can more easily see the target tissue in the prostate lymph node in the target pathological mask image.

[0129] In one embodiment, on the premise that the original pathological mask image includes the target tissue mask part, when the target pathological recognition result indicates that the prostate lymph node does not include the target tissue, pixel reduction is performed according to the target confidence level and the target tissue mask part to delete the target tissue mask information in the original pathological mask image. Among them, pixel reduction can be achieved by determining a reduction coefficient according to the target confidence level and generating multiple reduction pixels based on the pixel values of the pixels of the target tissue mask part (i.e., the original pixels appearing below), and specific details are not limited here.

[0130] In one embodiment, after step 903, the transitional pathological feature image and the first confidence level can be stored in a preset pathological image library, so as to increase the number of difficult samples in the target classification cluster, and further improve the accuracy of the target pathological mask image corresponding to the subsequent pathological image.

[0131] In one embodiment, step 902 can be implemented in the following manner: determining a corrected target weight according to the target confidence level, performing convolution on the target tissue mask part through a preset convolution kernel and the corrected target weight to generate target pixels corresponding to each pixel in the target tissue mask part, and replacing the pixels in the target tissue mask part with these target pixels. The corrected target weight determined by the target confidence level and the preset convolution kernel enable the target tissue mask part to enhance details through the target confidence level, thereby improving the masking effect of the target pathological mask image.

[0132] Refer to Figure 10 as shown in Figure 10 which shows Figure 9 sub-steps of a specific embodiment of step 902. In one embodiment, the target tissue mask part includes a plurality of original pixels, and step 902 may include the following steps.

[0133] Step 1001: For each original pixel, adjust the pixel value of the original pixel according to the target confidence level to obtain the target pixel value corresponding to the original pixel;

[0134] Step 1002: Generate a plurality of target pixels according to the plurality of target pixel values.

[0135] In one embodiment, step 1001 can be represented by the following formula (2).

[0136] (2),

[0137] In formula (2), can represent the pixel value of the target pixel;

[0138] can represent the pixel value of the original pixel;

[0139] can represent the target confidence level obtained in formula (1);

[0140] can represent a preset pixel correction coefficient, where the preset pixel correction coefficient can be 0.1, 0.2, 0.3, 0.4, 0.5, etc., and is not specifically limited;

[0141] It can represent a preset pixel value threshold. Herein, the preset pixel value threshold can be determined according to the attributes of the pixel values of the original pixels. When the pixel values of the original pixels have been subjected to interval conversion processing during storage, the preset pixel value threshold can correspondingly be 255; when the pixel values of the original pixels have not been subjected to interval conversion processing during storage, the preset pixel value threshold can correspondingly be 1, and no specific limitation is made here.

[0142] By correcting the pixel value of the original pixel with the target confidence level, the preset pixel correction coefficient, and the preset pixel value threshold, the pixel value of the target pixel can be obtained quickly, thereby quickly improving the generation efficiency of the target pixel; moreover, since the minimum value among the multiple second confidence levels corresponding to the target pathological image classification cluster is introduced during the process of correcting the first confidence level, the difference between the transitional pathological feature image and the most difficult sample in the target pathological image classification cluster is increased, thereby being able to improve the ability to correct the pixel value of the original pixel, and further enabling the target tissue mask information carried by the target pathological mask image to be more accurate and more obvious.

[0143] In one embodiment, the pathological image is obtained by acquiring a pathological section image of the prostate lymph node and performing a sliding window cropping process on the pathological section image. Among them, multiple pathological images are obtained by performing a sliding window cropping process on the pathological section image. After step 104, the histopathological recognition method for prostate lymph node metastasis may further include the following: performing pathological recognition on the next pathological image to output the target pathological mask image corresponding to the next pathological image. After obtaining the target pathological mask images corresponding to multiple pathological images, graphic merging is performed according to the target pathological mask images corresponding to the multiple pathological images to obtain the target pathological section mask image corresponding to the pathological section image, where the target pathological section mask image includes the target tissue main body mask part of the prostate lymph node. In addition, the image merging may include merging and edge smoothing processing, thereby improving the visual effect of the target tissue main body mask in the target pathological section mask image. By performing a sliding window cropping process on the pathological section image of the prostate lymph node, the feature amount of the pathological section image can be reduced, thereby reducing the difficulty of outputting the target pathological section mask image and improving the histopathological recognition efficiency of prostate lymph node metastasis.

[0144] Refer to Figure 11 As shown, the embodiment of the present application also discloses a histopathological recognition device for prostate lymph node metastasis. The histopathological recognition device 1100 for prostate lymph node metastasis can implement the histopathological recognition method in the previous embodiments. The histopathological recognition device 1100 includes:

[0145] An original pathological recognition module 1101, configured to perform pathological recognition on the pathological image of the prostate lymph node to obtain the first confidence level corresponding to the original pathological recognition result;

[0146] The original pathological mask processing module 1102 is used to perform mask processing on a pathological image to obtain an original pathological mask image;

[0147] The target confidence acquisition module 1103 is used to correct the first confidence according to a plurality of pathological example images in a preset pathological image library to obtain a target confidence;

[0148] The target pathological mask image output module 1104 is used to, when the original pathological mask image includes a target tissue mask part, perform pixel correction on the target tissue mask part according to the target confidence and output a target pathological mask image.

[0149] It should be noted that since the histopathology recognition device 1100 in this embodiment can implement the histopathology recognition method in the previous embodiment, the histopathology recognition device 1100 in this embodiment and the histopathology recognition method in the previous embodiment have the same technical principle and the same beneficial effects. To avoid repetition of content, it will not be elaborated here.

[0150] Refer to Figure 12 As shown, this application embodiment also discloses an electronic device. The electronic device 1200 includes:

[0151] At least one processor 1201;

[0152] At least one memory 1202, which is used to store at least one program;

[0153] When at least one program is executed by at least one processor 1201, the histopathology recognition method as described above is implemented.

[0154] This application embodiment also discloses a computer-readable storage medium, in which a computer program executable by a processor is stored. When the computer program executable by the processor is executed by the processor, it is used to implement the histopathology recognition method as described above.

[0155] This application embodiment also discloses a computer program product, including a computer program or computer instructions. The computer program or computer instructions are stored in a computer-readable storage medium. The processor of the electronic device reads the computer program or computer instructions from the computer-readable storage medium, and the processor executes the computer program or computer instructions, so that the electronic device executes the histopathology recognition method as described above.

[0156] In the description of this application and the above-mentioned drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0157] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (item) of the following" or its similar expression refers to any combination of these items, including any combination of single item (s) or plural item (s). For example, at least one (item) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0158] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be an indirect coupling or communication connection through some interfaces, devices, or units, and can be in electrical, mechanical, or other forms.

[0159] In the embodiments of the present application, the term "module" or "unit" refers to a computer program with a predetermined function or a part of a computer program, which works together with other related parts to achieve a predetermined goal, and can be fully or partially implemented by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Similarly, one processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of an overall module or unit that includes the functions of that module or unit.

[0160] The unit described as a separate component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0161] In addition, in each embodiment of the present application, the functional units can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0162] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several 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 each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc.

[0163] For the step numbers in the above method embodiments, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

Claims

1. A histopathological identification method for lymph node metastasis of prostate cancer, characterized in that, Including the following steps: Performing pathological recognition based on the pathological image of the prostate lymph nodes to obtain the first confidence corresponding to the original pathological recognition result, where feature extraction is performed on the pathological image to obtain a transitional pathological feature image; in addition, the original pathological recognition result is the recognition result finally output by pathological recognition, the original pathological recognition result is used to represent one or more target tissues on the pathological image, and the first confidence is the classification confidence corresponding to the target tissue; Performing masking processing on the pathological image to obtain an original pathological mask image; Performing cluster classification on the transitional pathological feature image according to multiple pathological image classification clusters in a preset pathological image library, and determining a target pathological image classification cluster from the multiple pathological image classification clusters, where the target pathological image classification cluster includes multiple pathological example images; Correcting the first confidence according to the multiple pathological example images in the target pathological image classification cluster to obtain a target confidence; When the original pathological mask image includes a target tissue mask part, determining a target pathological recognition result according to the target confidence, where the target tissue mask part includes multiple original pixels; When the target pathological recognition result indicates that the prostate lymph nodes include the target tissue, for each of the original pixels, adjusting the pixel value of the original pixel according to the target confidence to obtain a target pixel value corresponding to the original pixel; Generating multiple target pixels corresponding to the multiple target pixel values; Performing pixel replacement on the target tissue mask part according to the multiple target pixels and outputting a target pathological mask image.

2. The method according to claim 1, characterized in that, The preset pathological image library includes pathological image classification labels corresponding to each of the pathological image classification clusters, and a second confidence corresponding to each of the pathological example images, where the second confidence is obtained by performing pathological recognition on the pathological example image and is the recognition confidence corresponding to the pathological image classification label; The correcting the first confidence according to the multiple pathological example images in the target pathological image classification cluster to obtain the target confidence includes: Determining a target preset correction offset value corresponding to the target pathological image classification cluster from multiple preset correction offset values according to the pathological image classification label of the target pathological image classification cluster, where the multiple preset correction offset values correspond to the multiple pathological image classification clusters one by one; Correcting the first confidence according to the multiple second confidences corresponding to the target pathological image classification cluster and the target preset correction offset value to obtain the target confidence.

3. The method according to claim 2, characterized in that, The correcting the first confidence according to the multiple second confidences corresponding to the target pathological image classification cluster and the target preset correction offset value to obtain the target confidence includes: Determining a third confidence and a fourth confidence among the multiple second confidences corresponding to the target pathological image classification cluster; Correcting the first confidence according to the third confidence, the fourth confidence and the target preset correction offset value to obtain the target confidence.

4. The method according to claim 1, wherein The pathological image is obtained by acquiring a pathological section image of a prostate lymph node and performing a sliding window cropping process on the pathological section image. Among them, a plurality of the pathological images are obtained by performing the sliding window cropping process on the pathological section image; The method further includes: Performing pathological recognition on the next pathological image to output the target pathological mask image corresponding to the next pathological image; After obtaining the target pathological mask images corresponding to the plurality of pathological images, performing graphic merging according to the target pathological mask images corresponding to the plurality of pathological images to obtain the target pathological section mask image corresponding to the pathological section image, where the target pathological section mask image includes the target tissue main body mask part of the prostate lymph node.

5. A histopathological recognition device for lymph node metastasis of prostate cancer, characterized in that, It includes: An original pathological recognition module, configured to perform pathological recognition according to a pathological image of a prostate lymph node to obtain a first confidence level corresponding to an original pathological recognition result, where it includes extracting features from the pathological image to obtain a transitional pathological feature image; in addition, the original pathological recognition result is the recognition result finally output by pathological recognition, the original pathological recognition result is used to represent one or more target tissues on the pathological image, and the first confidence level is the classification confidence level corresponding to the target tissue; An original pathological mask processing module, configured to perform mask processing on the pathological image to obtain an original pathological mask image; A target confidence level acquisition module, configured to perform cluster classification on the transitional pathological feature image according to a plurality of pathological image classification clusters in a preset pathological image library, and determine a target pathological image classification cluster from the plurality of pathological image classification clusters, where the target pathological image classification cluster includes a plurality of pathological example images; Correcting the first confidence level according to the plurality of pathological example images in the target pathological image classification cluster to obtain a target confidence level; A target pathological mask image output module, configured to, when the original pathological mask image includes a target tissue mask part, determine a target pathological recognition result according to the target confidence level, where the target tissue mask part includes a plurality of original pixels; when the target pathological recognition result indicates that the prostate lymph node includes the target tissue, for each of the original pixels, adjusting the pixel value of the original pixel according to the target confidence level to obtain a target pixel value corresponding to the original pixel; generating a plurality of target pixels according to the plurality of target pixel values; replacing the pixels of the target tissue mask part with the plurality of target pixels, and outputting a target pathological mask image.

6. An electronic device, characterized in that, It includes: At least one processor; At least one memory, configured to store at least one program; When at least one of the at least one program is executed by the at least one processor, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that, Wherein there is stored a computer program executable by a processor, and when the computer program executable by the processor is executed by the processor, it is used to implement the method according to any one of claims 1 to 4.

Images