Leukocyte classification method and system, terminal and storage medium

Abstract

The invention provides a leukocyte classification method and system, a terminal and a storage medium, and relates to the technical field of medical image processing, in particular to a leukocyte classification method, which comprises the following steps: acquiring a leukocyte image sample carrying a classification tag; performing preprocessing and data enhancement on the leukocyte image sample to obtain a processed training sample; replacing a convolution kernel with a relatively large size in the NASNet network with a convolution kernel with a relatively small size to obtain an improved NASNet network, and based on the improved NASNet network, constructing a classification model which takes the leukocyte image as an input and takes the probability value of the leukocyte image belonging to each classification label as an output; training the classification model by using the training sample to obtain a trained classification model; and carrying out leukocyte classification by using the trained classification model. The method can improve the accuracy of leukocyte classification.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a white blood cell classification method, system, terminal and storage medium. Background Art

[0002] For routine blood tests, the five-category blood analyzers are widely used in clinical laboratories of hospitals at all levels, especially in the fields of hematology, infectious diseases, and oncology. These instruments usually use the electrical impedance method combined with flow cytometry and fluorescent staining technology. In addition to counting white blood cells, red blood cells, and platelets, the five-category blood analyzers can also classify white blood cells in more detail, subdividing white blood cells into neutrophils, lymphocytes, monocytes, basophils, and eosinophils. Among them, flow cytometry can make a rough estimate of the normal white blood cell category, but this classification method is not accurate enough.

[0003] Therefore, how to improve the accuracy of white blood cell classification is a technical problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0004] To solve the above technical problems, the present application provides a leukocyte classification method, which can improve the accuracy of leukocyte classification. The present application also provides a leukocyte classification system, a terminal and a storage medium, which have the same technical effects.

[0005] The first object of the present application is to provide a method for classifying leukocytes.

[0006] The above-mentioned application objective 1 of the present application is achieved through the following technical solutions: A method for classifying leukocytes, comprising: Obtaining a sample of a white blood cell image carrying a classification label; Preprocessing and data enhancement are performed on the white blood cell image samples to obtain processed training samples; Replacing the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network, and based on the improved NASNet network, constructing a classification model with the white blood cell image as input and the probability value of the white blood cell image belonging to each of the classification labels as output; Using the training samples, training the classification model to obtain a trained classification model; The trained classification model is used to classify white blood cells.

[0007] Preferably, in the white blood cell classification method, replacing a larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network comprises: The 5x5 convolution kernel in the NASNet network is replaced by a 3x3 convolution kernel to obtain the improved NASNet network.

[0008] Preferably, the leukocyte classification method further comprises, before inputting the leukocyte image into the classification model and / or the trained classification model: Resize the WBC image to the preset resolution size.

[0009] Preferably, in the white blood cell classification method, replacing a larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network comprises: The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels, and the number of pooling layers in the NASNet network is reduced to obtain an improved NASNet network.

[0010] Preferably, in the white blood cell classification method, the preprocessing method includes Laplacian operator and bilateral filtering; The data enhancement method includes image rotation, flipping and affine transformation.

[0011] Preferably, in the white blood cell classification method, using the training samples to train the classification model to obtain a trained classification model includes: Dividing the training samples into a training set, a validation set and a test set; The classification model is trained by using the training set, and the parameters are tuned by using the early stopping mechanism and the grid search method by using the validation set to obtain a trained intermediate model; Using the test set, performing performance evaluation on the intermediate model to obtain an evaluation index; Determine whether the evaluation index meets the preset requirements, and if so, use the intermediate model as the trained classification model.

[0012] Preferably, in the white blood cell classification method, the white blood cell classification using the trained classification model comprises: Acquire an image of white blood cells to be classified; Inputting the to-be-classified leukocyte image into the trained classification model to obtain a probability value of the to-be-classified leukocyte image belonging to each of the classification labels; The classification label corresponding to the maximum probability value is taken as the category of the white blood cell image to be classified.

[0013] The second object of the present application is to provide a leukocyte classification system.

[0014] The second application objective of the present application is achieved through the following technical solutions: A leukocyte classification system comprising: An acquisition unit, used for acquiring a white blood cell image sample carrying a classification label; A processing unit, used for preprocessing and data enhancement of the white blood cell image sample to obtain a processed training sample; A construction unit is used to replace the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network, and based on the improved NASNet network, a classification model is constructed with the white blood cell image as input and the probability value of the white blood cell image belonging to each of the classification labels as output; A training unit, used to train the classification model using the training samples to obtain a trained classification model; The classification unit is used to classify white blood cells using the trained classification model.

[0015] The third object of the present application is to provide a leukocyte classification terminal.

[0016] The third application objective of this application is achieved through the following technical solutions: A leukocyte classification terminal comprises: a storage medium and a processor; The storage medium stores computer-executable instructions; The processor executes the computer-executable instructions stored in the storage medium to implement the white blood cell classification method as described above.

[0017] The fourth object of this application is to provide a computer-readable storage medium.

[0018] The fourth application objective of this application is achieved through the following technical solutions: A computer-readable storage medium stores computer-executable instructions, which are used to implement the above-mentioned leukocyte classification method when executed by a processor.

[0019] The above technical scheme obtains white blood cell image samples carrying classification labels; preprocesses and data enhances the white blood cell image samples to obtain processed training samples; replaces the larger convolution kernels in the NASNet network with smaller convolution kernels to obtain an improved NASNet network, and based on the improved NASNet network, constructs a classification model that takes the white blood cell images as input and outputs the probability values ​​of the white blood cell images belonging to each classification label; uses the training samples to train the classification model to obtain a trained classification model; and uses the trained classification model to classify white blood cells.

[0020] Compared with the existing flow cytometry methods, the above technical solution adopts a deep learning method to achieve accurate classification of white blood cell images. In addition, by replacing the larger convolution kernels in the NASNet network with smaller convolution kernels, it can better adapt to the local features of white blood cell images, capture the fine textures and details in white blood cell images, and improve the classification performance of the classification model.

[0021] In summary, the above technical solution can improve the accuracy of white blood cell classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 A schematic diagram of a process of a leukocyte classification method in an embodiment of the present application; Figure 2 This is a schematic diagram of the structure of a leukocyte classification system in an embodiment of the present application; Figure 3 This is a schematic diagram of the structure of a white blood cell classification terminal in an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] In the embodiments provided in the present application, it should be understood that the disclosed methods and systems can be implemented in other ways. The system embodiments described below are merely schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as: multiple modules or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0026] It should be understood that the use of "system", "device", "unit" and / or "module" in this application is only a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the word can be replaced by other expressions.

[0027] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "multiple" and "several" mean two or more, unless otherwise clearly and specifically defined.

[0028] If a flow chart is used in the present application, the flow chart is used to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, each step can be processed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or a certain step or several steps of operations can be removed from these processes.

[0029] It should also be noted that, in this article, terms such as "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that an article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such articles or devices. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the article or device including the above elements.

[0030] The embodiments of the present application are written in a progressive manner.

[0031] like Figure 1 As shown, the present application embodiment provides a method for classifying leukocytes, comprising: S101. Obtaining a white blood cell image sample carrying a classification label; In S101, specifically, manually annotated white blood cell images carrying classification labels can be obtained as sample data for subsequent model training; wherein the classification labels can include five types of labels: neutrophils, lymphocytes, monocytes, basophils, and eosinophils.

[0032] S102. Preprocessing and data enhancement are performed on the white blood cell image samples to obtain processed training samples; In S102, specifically, the white blood cell image samples can be preprocessed to obtain preprocessed image samples, and then the preprocessed image samples can be data enhanced to obtain processed training samples; wherein, the quality of the white blood cell images can be improved by preprocessing, and the capacity of the data set can be increased by data enhancement, thereby ensuring that the number of data for each classification label is at the same level, thereby avoiding problems caused by data imbalance.

[0033] In some embodiments, the preprocessing method includes Laplace operator and bilateral filtering; specifically, the Laplace operator is used to calculate the white blood cell image sample to obtain the image sample after Laplace transformation; then the bilateral filtering is used to denoise the image sample after Laplace transformation to obtain the preprocessed image sample.

[0034] Among them, the Laplace operator is a second-order differential operator, which is used to calculate the second-order derivative of the grayscale gradient of the image, characterizing the steepness of the local grayscale change of the image. Through the Laplace operator calculation, the edge and detail information in the white blood cell image can be enhanced, which is helpful for subsequent feature extraction and analysis. Bilateral filtering is a nonlinear filtering method that considers both the spatial distance and color / grayscale similarity between pixels. Bilateral filtering can effectively remove image noise and keep the image edges clear, which is very beneficial for image enhancement and detail protection.

[0035] In other embodiments, the data enhancement method includes image rotation, flipping and affine transformation; specifically, the preprocessed image samples are randomly subjected to image rotation, flipping and affine transformation operations to obtain processed training samples. Through data enhancement, the number of samples used for training and testing models can be increased, which can alleviate model overfitting and improve the generalization ability of the training model.

[0036] S103. Replace the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network. Based on the improved NASNet network, construct a classification model that takes the white blood cell image as input and outputs the probability value of the white blood cell image belonging to each classification label; In S103, specifically, the architecture of the NASNet network consists of two main parts: normal cells and reduction cells. Normal cells are used to maintain the dimension of the feature map, while reduction cells reduce the height and width of the feature map by half. NASNet predicts the structure of these neurons through a recurrent neural network (RNN) controller, which recursively predicts the configuration of the convolutional unit, including selecting hidden states, determining the operation mode, and combining operations.

[0037] Considering that the choice of convolution kernel size has a significant impact on the performance of the image classification model, in this step, in order to better adapt to the classification task of white blood cell images, the larger convolution kernels in the NASNet network are replaced with smaller convolution kernels, so as to better adapt to the local features of white blood cell images, capture the fine texture and details in white blood cell images, and improve the classification performance of the model; based on the improved NASNet network, a classification model is constructed with white blood cell images as input and the probability value of the white blood cell images belonging to each classification label as output, and accurate classification of white blood cell images is achieved through model training.

[0038] In some embodiments, one implementation of the step of replacing a larger convolution kernel in a NASNet network with a smaller convolution kernel to obtain an improved NASNet network specifically includes: replacing a 5x5 convolution kernel in a NASNet network with a 3x3 convolution kernel to obtain an improved NASNet network.

[0039] Among them, smaller convolution kernels (such as 3x3 convolution kernels) have local receptive fields and can capture fine textures and details in images. For white blood cell image classification, which is an image classification task in which local features are important and requires fast inference, 3x3 convolution kernels are a better choice compared to larger convolution kernels (such as 5x5 convolution kernels). Accordingly, the classification model built based on the improved NASNet network has better classification performance.

[0040] S104. Using the training samples, the classification model is trained to obtain a trained classification model; In S104, specifically, based on the existing model training method, the classification model can be iteratively trained using training samples, and when a preset iteration termination condition is met, a trained classification model is obtained.

[0041] In some embodiments, one implementation of this step specifically includes: S1041. Divide the training samples into a training set, a validation set, and a test set; In S1041, specifically, the training samples can be divided into a training set, a validation set, and a test set according to a preset ratio, such as a ratio of 6:2:2. The training set is used to train the model, the validation set is used to adjust the model parameters and evaluate the model performance, and the test set is used to finally evaluate the generalization ability of the model to ensure that the model performs well on new data that has not been seen.

[0042] S1042. The classification model is trained using the training set, and the parameters are tuned using the early stopping mechanism and the grid search method using the validation set to obtain a trained intermediate model; In S1042, specifically, the entire training process first performs forward propagation. In each iteration of model training, the image data in the training set is forward propagated and loss is calculated through the neural network, and then the model parameters are optimized through back propagation and stochastic gradient descent (SGD) algorithm so that the model can fit the training data as much as possible. During the training process, by monitoring the performance indicators on the validation set, the early stopping mechanism and grid search method are adopted to tune the hyperparameters of the model, find the optimal model configuration, prevent overfitting, and obtain a trained intermediate model.

[0043] S1043. Using the test set, evaluate the performance of the intermediate model to obtain evaluation indicators; In S1043, specifically, the test set is used to finally test the model performance. The evaluation indicators of the intermediate model can be calculated based on the model classification results of the test set, where the evaluation indicators include recall rate (Recall), precision (Precision), etc., and this application does not impose specific restrictions on this.

[0044] S1044. Determine whether the evaluation index meets the preset requirements. If so, use the intermediate model as the trained classification model.

[0045] In S1044, specifically, the value of each evaluation indicator can be compared with a preset threshold range. If the value of each evaluation indicator is within the preset threshold range, it can be considered that the evaluation indicator meets the preset requirements, and the intermediate model can be used as a trained classification model.

[0046] S105. Use the trained classification model to classify white blood cells.

[0047] In S105, specifically, after the trained classification model is obtained, the trained classification model can be used to classify the white blood cells in the white blood cell image to be classified.

[0048] In some embodiments, one implementation of this step specifically includes: obtaining a white blood cell image to be classified; inputting the white blood cell image to be classified into a trained classification model to obtain a probability value that the white blood cell image to be classified belongs to each classification label; taking the classification label corresponding to the maximum probability value as the category of the white blood cell image to be classified. Among them, the white blood cell image to be classified can be acquired through existing equipment, and this application does not impose specific restrictions on this. In other embodiments, the white blood cell image to be classified can also be pre-processed and then input into a trained classification model. The pre-processing method can refer to step S102, and this application is not limited to this.

[0049] In the above embodiment, a white blood cell image sample carrying a classification label is obtained; the white blood cell image sample is preprocessed and data enhanced to obtain a processed training sample; a larger convolution kernel in the NASNet network is replaced with a smaller convolution kernel to obtain an improved NASNet network; based on the improved NASNet network, a classification model is constructed with the white blood cell image as input and the probability value of the white blood cell image belonging to each classification label as output; the classification model is trained using the training sample to obtain a trained classification model; and the white blood cells are classified using the trained classification model. Compared with the existing flow cytometry method, the above embodiment adopts a deep learning method to achieve accurate classification of white blood cell images. In addition, by replacing the larger convolution kernel in the NASNet network with a smaller convolution kernel, it can better adapt to the local features of the white blood cell image, capture the fine texture and details in the white blood cell image, and improve the classification performance of the classification model. In summary, the above embodiment can improve the accuracy of white blood cell classification.

[0050] In other embodiments of the present application, in the above-mentioned white blood cell classification method, before the white blood cell image is input into the classification model and / or the trained classification model, it also includes: adjusting the white blood cell image to a preset resolution size.

[0051] Accordingly, one implementation method of the steps of replacing the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain the improved NASNet network specifically includes: The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels, and the number of pooling layers in the NASNet network is reduced to obtain an improved NASNet network.

[0052] Specifically, for white blood cell images, due to their relatively small size (the largest is about 50 pixels wide and high), directly applying the standard NASNet network may not be the best choice; in this embodiment, all white blood cell images are first resized to a preset resolution size, such as 64x64, and the size is slightly enlarged and fixed to facilitate input into the classification model.

[0053] In addition, considering the small size of white blood cell images, excessive downsampling will cause the resolution of the feature map to decrease rapidly, and important detail information may be lost. In this embodiment, on the basis of replacing the larger convolution kernels in the NASNet network with smaller convolution kernels, the number of pooling layers in the NASNet network is reduced to maintain more spatial resolution, which is conducive to improving the classification accuracy of the classification model. For example, the number of pooling layers that need to be reduced can be pre-set for different image input sizes. After determining the image input size, the set number of pooling layers will be reduced accordingly, and when the last layer of the fully connected layer used for decision-making in the input classification model is used, the image size should be (B, C, 16, 16), where B represents batch size, C represents channels, which is the number of channels after the final convolution, and 16 and 16 represent the length and width of the image, respectively.

[0054] like Figure 2 As shown, in another embodiment of the present application, a leukocyte classification system is also provided, comprising: An acquisition unit 10, used to acquire a white blood cell image sample carrying a classification label; A processing unit 11 is used to perform preprocessing and data enhancement on the white blood cell image samples to obtain processed training samples; A construction unit 12 is used to replace the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network, and based on the improved NASNet network, a classification model is constructed with the white blood cell image as input and the probability value of the white blood cell image belonging to each classification label as output; The training unit 13 is used to train the classification model using the training samples to obtain a trained classification model; The classification unit 14 is used to classify white blood cells using the trained classification model.

[0055] In other embodiments of the present application, the construction unit 12, when replacing the larger convolution kernel in the NASNet network with the smaller convolution kernel to obtain the improved NASNet network, is specifically used to: The 5x5 convolution kernel in the NASNet network is replaced by a 3x3 convolution kernel to obtain the improved NASNet network.

[0056] In other embodiments of the present application, the above-mentioned white blood cell classification system further includes an adjustment unit, wherein the adjustment unit is used to adjust the white blood cell image to a preset resolution size before inputting the white blood cell image into the classification model and / or the trained classification model.

[0057] In other embodiments of the present application, the construction unit 12, when replacing the larger convolution kernel in the NASNet network with the smaller convolution kernel to obtain the improved NASNet network, is specifically used to: The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels, and the number of pooling layers in the NASNet network is reduced to obtain an improved NASNet network.

[0058] In other embodiments of the present application, the preprocessing methods include Laplacian operator and bilateral filtering; the data enhancement methods include image rotation, flipping and affine transformation.

[0059] In other embodiments of the present application, the training unit 13, when executing the training sample to train the classification model to obtain the trained classification model, is specifically used to: Divide the training samples into training set, validation set and test set; The classification model is trained through the training set, and the parameters are tuned through the validation set using the early stopping mechanism and grid search method to obtain a trained intermediate model; Use the test set to evaluate the performance of the intermediate model and obtain the evaluation index; Determine whether the evaluation indicators meet the preset requirements. If so, use the intermediate model as the trained classification model.

[0060] In other embodiments of the present application, the classification unit 14, when executing the classification of white blood cells using the trained classification model, is specifically used to: Acquire an image of white blood cells to be classified; Input the white blood cell image to be classified into the trained classification model to obtain the probability value of the white blood cell image to be classified belonging to each classification label; The classification label corresponding to the maximum probability value is taken as the category of the white blood cell image to be classified.

[0061] like Figure 3 As shown, in another embodiment of the present application, a leukocyte classification terminal is also provided, comprising: a storage medium 20 and a processor 21; The storage medium 20 stores computer-executable instructions; The processor 21 executes the computer-executable instructions stored in the storage medium 20 to implement the white blood cell classification method as described above.

[0062] Among them, the processor 21 may include one or more processing cores. The processor 21 executes various functions and processes data of the present application by running or executing instructions, programs, code sets or instruction sets stored in the storage medium 20, calling data stored in the storage medium 20. The processor 21 can be at least one of a special application integrated circuit, a digital signal processor, a digital signal processing device, a programmable logic device, a field programmable gate array, a central processing unit, a controller, a microcontroller and a microprocessor. It can be understood that for different devices, the electronic device used to implement the functions of the above-mentioned processor 21 can also be other.

[0063] The storage medium 20 may be used to store instructions, programs, codes, code sets or instruction sets. The storage medium 20 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function, and instructions for implementing the above-mentioned leukocyte classification method, etc.; the data storage area may store data involved in the above-mentioned leukocyte classification method, etc.

[0064] In another embodiment of the present application, a computer-readable storage medium is provided, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the white blood cell classification method as described above.

[0065] The computer-readable storage medium may be a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, or a CD, etc., which may store program codes.

[0066] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for classifying leukocytes, characterized in that: include: Obtaining a sample of a white blood cell image carrying a classification label; Preprocessing and data enhancement are performed on the white blood cell image samples to obtain processed training samples; Replacing the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network, and based on the improved NASNet network, constructing a classification model with the white blood cell image as input and the probability value of the white blood cell image belonging to each of the classification labels as output; Using the training samples, training the classification model to obtain a trained classification model; The trained classification model is used to classify white blood cells.

2. The method according to claim 1, characterized in that The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels to obtain an improved NASNet network, including: The 5x5 convolution kernel in the NASNet network is replaced by a 3x3 convolution kernel to obtain the improved NASNet network.

3. The method according to claim 1, characterized in that Before inputting the white blood cell image into the classification model and / or the trained classification model, the method further includes: Resize the WBC image to the preset resolution size.

4. The method as claimed in claim 3, characterized in that The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels to obtain an improved NASNet network, including: The larger convolution kernels in the NASNet network are replaced with smaller convolution kernels, and the number of pooling layers in the NASNet network is reduced to obtain an improved NASNet network.

5. The method according to claim 1, characterized in that The preprocessing method includes Laplacian operator and bilateral filtering; The data enhancement method includes image rotation, flipping and affine transformation.

6. The method according to claim 1, characterized in that The step of training the classification model using the training samples to obtain a trained classification model includes: Dividing the training samples into a training set, a validation set and a test set; The classification model is trained by using the training set, and the parameters are tuned by using the early stopping mechanism and the grid search method by using the validation set to obtain a trained intermediate model; Using the test set, performing performance evaluation on the intermediate model to obtain an evaluation index; Determine whether the evaluation index meets the preset requirements, and if so, use the intermediate model as the trained classification model.

7. The method according to claim 1, characterized in that The step of using the trained classification model to classify white blood cells includes: Acquire an image of white blood cells to be classified; Inputting the to-be-classified leukocyte image into the trained classification model to obtain a probability value of the to-be-classified leukocyte image belonging to each of the classification labels; The classification label corresponding to the maximum probability value is taken as the category of the white blood cell image to be classified.

8. A leukocyte classification system, characterized in that: include: An acquisition unit, used for acquiring a white blood cell image sample carrying a classification label; A processing unit, used for preprocessing and data enhancement of the white blood cell image sample to obtain a processed training sample; A construction unit is used to replace the larger convolution kernel in the NASNet network with a smaller convolution kernel to obtain an improved NASNet network, and based on the improved NASNet network, a classification model is constructed with the white blood cell image as input and the probability value of the white blood cell image belonging to each of the classification labels as output; A training unit, used to train the classification model using the training samples to obtain a trained classification model; The classification unit is used to classify white blood cells using the trained classification model.

9. A leukocyte classification terminal, characterized in that: include: Storage media and processors; The storage medium stores computer-executable instructions; The processor executes the computer-executable instructions stored in the storage medium to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images