Method, System, Device and Medium for Counting Blood Cell Quantity Based on Image Recognition

Abstract

The present application discloses a method, system, device and medium for counting the number of blood cells based on image recognition, which relates to the technical field of image data processing. The method includes: recognizing a blood smear image based on a recognition model, determining and labeling the cell type and accurate probability of each cell image; generating a probability distribution map based on the position distribution law of various types of blood cells on the blood smear; obtaining the cell type, accurate probability and position coordinates of each cell image, combining the probability distribution map to assign a reliability weight to the accurate probability, generating a type determination reliability value for each cell image and comparing it with a set threshold, determining the cell type according to the comparison result and labeling it, and counting and outputting the number of various types of blood cells according to the labeling. The above solution combines probability distribution to calculate the blood cell morphology and quantity of cell overlapping images while recognizing blood cell images based on AI, improves the accuracy of blood cell morphology identification and counting, and provides reliable data reference for doctor decision-making.

Description

Technical Field

[0001] This application relates to the technical field of image data processing, and in particular to a method, system, device and medium for counting the number of blood cells based on image recognition. Background Art

[0002] In the field of medical examination, the morphological examination of blood cells is crucial for the diagnosis of various diseases, such as leukemia, anemia, infection, etc. Traditional morphological diagnosis of blood cells mainly relies on laboratory technicians to manually observe blood smears through a microscope, identify the morphology, quantity and abnormal characteristics of various blood cells by experience, and then write a diagnostic report.

[0003] Currently, more and more medical examination institutions and medical device manufacturers have introduced AI image analysis technology into the field of blood cell morphological diagnosis. The AI analysis module with a built-in blood cell recognition model automatically performs image recognition on the blood cell images of blood smears, can accurately identify various blood cells and their subtypes such as red blood cells, white blood cells, and platelets, and at the same time can accurately detect morphological abnormalities of cells, such as uneven size, abnormal shape, nuclear-cytoplasmic changes, etc., and outputs an analysis result containing detailed information such as cell type, quantity, and abnormal characteristics. Then, according to the corresponding relationship between blood cell morphological changes and common diseases, an auxiliary diagnostic report is generated for doctors to refer to for decision-making.

[0004] As more and more blood cell image materials are fed into the training model, the recognition accuracy of the blood cell recognition model has reached over 95%. Currently, the main factor affecting the identification and counting of blood cell morphology is that cells on the blood smear will overlap, and the morphology of overlapping blood cells is difficult to be effectively recognized by the recognition model. In fact, for the case where different types of blood cells completely overlap and cover, even manual microscopy is difficult to accurately judge and distinguish the cell morphology and quantity constituting the overlapping image, and the use of manual microscopy is inefficient, thus having an adverse impact on the accurate identification and counting of blood cell morphology. Summary of the Invention

[0005] To solve the problems of insufficient accuracy and low efficiency in the existing blood cell morphology diagnosis, one objective of this application is to provide a method for counting the number of blood cells based on image recognition. It recognizes blood cell images based on artificial intelligence technology, and combines probability distribution to deduce the morphology and number of blood cells in the overlapping cell images, further improving the accuracy and efficiency of blood cell morphology identification and counting, and providing reliable data reference for doctors' decision-making. To implement the above method for counting the number of blood cells based on image recognition, a second objective of this application is to provide a system for counting the number of blood cells based on image recognition, which not only optimizes the recognition software, but also fits the hardware configuration with the method, further improving the accuracy of blood cell counting. A third objective is to provide a device for counting the number of blood cells based on image recognition. Finally, it proposes to protect a computer-readable storage medium, on which a computer program module for implementing the method for counting the number of blood cells based on image recognition is loaded. The specific solutions are as follows:

[0006] A method for counting the number of blood cells based on image recognition, comprising:

[0007] Recognize the blood smear image based on a recognition model, and determine and label the cell type and accurate probability corresponding to each cell image therein; the production of the above blood smear is automatically completed by a smearer, and the same blood smear production parameters are used when making the blood smear;

[0008] Obtain and generate a probability distribution map representing the probability of different types of blood cells appearing at different positions on the blood smear based on the position distribution law of different types of blood cells on the blood smear;

[0009] Obtain the cell type and accurate probability corresponding to each cell image after determination, as well as the position coordinates of each cell image, and assign a reliability weight to the accurate probability in combination with the probability distribution map to generate a type determination reliability value corresponding to each cell image;

[0010] Compare the type determination reliability value corresponding to each cell image with a set threshold:

[0011] If the type determination reliability value corresponding to the cell image exceeds the set threshold, it is determined as a definite sample and its cell type is labeled;

[0012] If the type determination reliability value corresponding to the cell image does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and number of the above sample to be determined are confirmed and labeled by manual and / or extended determination steps;

[0013] Based on the annotation information of each cell image in the blood smear image, count and output the number of each type of blood cell.

[0014] Through the above technical solution, first, the blood smear image is recognized based on artificial intelligence technology to determine the cell types corresponding to each cell image therein, and then the determination accuracy is evaluated, and the corresponding accurate probability is generated by association. Then, according to the distribution law of each type of cell on the blood smear, a reliability weight is assigned to the above accurate probability, thereby making the result of cell type determination more accurate. For some cell images with low determination accuracy, the accuracy of the determination result can be improved through weighted calculation, and finally the accuracy and efficiency of blood cell morphology identification and counting are improved, providing reliable data reference for doctors' diagnosis and treatment decisions.

[0015] Further, the cell images include single cell images and cell overlapping images;

[0016] In the recognition model, the association relationships between single cell images and cell types, and between cell overlapping images and the combinations of cell types and quantities that make up the above cell overlapping images are stored as a first relational database;

[0017] In the blood cell quantity counting method, the cell types corresponding to each cell image are determined and labeled as multiple, and each cell type generated by the determination corresponds to an accurate probability;

[0018] Determining and labeling the cell types and accurate probabilities corresponding to each cell image includes:

[0019] Obtain single cell images and cell overlapping images in the blood smear image, search for and match cell images similar to the above single cell images and cell overlapping images based on the first relational database, obtain the corresponding cell types or combinations of cell types and quantities, and label accurate probabilities for each matching result according to the similarity degree between cell images.

[0020] Through the above technical solution, each cell image in the blood smear image corresponds to a cell category respectively, and each determined cell category corresponds to an accurate probability, which is convenient for subsequent weighted calculation; at the same time, the cell overlapping image also corresponds to the combination of cell type and quantity, providing a primary determination basis for subsequent extended determination steps.

[0021] Further, the blood cell quantity counting method further includes:

[0022] Establish the association relationship among the overlapping probabilities of each type of cell, the cell overlapping positions, and the blood smear production parameters, and store it as a second relational database;

[0023] The extended determination step includes:

[0024] Obtain cell overlapping images, obtain multiple corresponding combinations of cell types and quantities through the first relational database, and configure a first accurate probability value for each combination;

[0025] Obtain the position coordinates of the cell overlapping images, configure the second accurate probability value for each combination of the cell types and quantities according to the second relational database and the probability distribution diagram, and configure the third accurate probability value for the cell types in the combination;

[0026] Obtain the blood smear production parameter data corresponding to the current blood smear image, and configure the fourth accurate probability value for each combination of the cell types and quantities according to the second relational database;

[0027] Assign reliable weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, and the fourth accurate probability value, and calculate and obtain the reliable values of the cell types, quantities, and quantity categories corresponding to each cell overlapping image according to the set algorithm;

[0028] Compare the above-mentioned reliable quantity category judgment value with a set threshold value:

[0029] If the reliable quantity category judgment value corresponding to the cell overlapping image exceeds the set threshold value, it is determined as a definite sample and its corresponding cell type and quantity are marked;

[0030] If the reliable quantity category judgment value corresponding to the cell overlapping image does not exceed the set threshold value, it is determined as a sample to be determined, and the cell type and quantity of the above-mentioned sample to be determined are confirmed manually and marked;

[0031] Wherein, the blood smear production parameters include the included angle between the spreading slide and the glass slide, the moving speed of the spreading slide, and the blood volume.

[0032] Through the above technical solution, introducing the influencing factors that affect the cell overlapping probability and the cell coincidence position into the determination of the cell type of the sample to be determined can effectively improve the determination accuracy of the cell type corresponding to the sample to be determined, and provide reliable data support for the doctor's subsequent diagnosis.

[0033] Furthermore, the blood cell quantity counting method further includes:

[0034] Divide all or part of the area of the blood smear image into multiple continuous and equal-area statistical regions;

[0035] Count the quantities of various types of cells included in the definite samples in each statistical region, and calculate the quantity proportion of various types of cells in the statistical region;

[0036] Fit and generate a trend curve of the above ratio changing with the statistical region according to the ratio of the quantity proportion of the set type of cells in each statistical region;

[0037] Obtain the combination of the cell types and quantities corresponding to the cell overlapping images in the statistical region, and disassemble the above combination of the cell types and quantities into the estimated quantities of various types of cells;

[0038] Add the estimated quantity of each type of cell to the statistical data of the current statistical area, and calculate the ratio of the proportion of each type of cell in the current statistical area before and after adding the above-mentioned estimated quantity;

[0039] Configure a fifth accuracy probability value for each of the cell type and quantity combinations according to the deviation degree of the above-mentioned proportion ratio from the trend curve;

[0040] Assign weight values to the first accuracy probability value, the second accuracy probability value, the third accuracy probability value, the fourth accuracy probability value and the fifth accuracy probability value, and calculate and obtain the cell types and quantities contained in the cell overlapping image according to the set algorithm;

[0041] Among them, the standard for dividing the statistical area is: the sum of the differences in the number of samples to be determined in each adjacent statistical area is the largest.

[0042] The distribution of various types of blood cells on the blood smear is usually uniformly and continuously changing, that is, the ratio of the proportion of each type of cell in each statistical area should be a smooth curve. Through the above technical solution, when the quantities of various types of cells inferred from the self-uncertain samples, that is, the cell overlapping images, are added to the existing quantities of various types of cells, if the ratio of the proportion of each type of cell changes too much, that is, the proportion ratio in the current statistical area deviates greatly from the trend curve, it can be inferred that the quantities of various types of cells inferred do not satisfy the probability distribution law in statistics, so the reliability of the inference is greatly reduced, and vice versa, the reliability increases; through the above technical solution, even if some cell images in the blood smear image are difficult to determine their original cell composition due to cell overlap for manual or AI image recognition, it is also possible to assign corresponding reliability weight values to various different cell types and quantity combinations through probability distribution, thereby further improving the reliability of cell type and quantity statistics.

[0043] Further, the cell quantity counting method further includes:

[0044] Receive and respond to an external image review request, and query the target cell type and perform enhancement or fading processing on the set cell image based on the annotation information of each cell image in the blood smear image.

[0045] Through the above technical solution, the inspector can quickly review the quantity and morphology of the set type of blood cells, and can also quickly obtain a judgment conclusion for some uncertain samples, improving the efficiency of the entire blood cell counting.

[0046] Further, the recognition of the blood smear image by the recognition model further includes an image preprocessing step: performing noise reduction, contrast enhancement, and gray correction operations on the obtained blood smear image to remove the noise interference generated during the image acquisition process, highlighting the morphological details of blood cells, and then outputting the preprocessed blood smear image to the recognition model.

[0047] Through the above technical solution, the image features of the blood smear image can be efficiently and accurately extracted, which is convenient for improving the recognition accuracy of the later recognition model and the reliability of the counting result.

[0048] A blood cell count system based on image recognition includes:

[0049] A data storage unit configured to obtain and store the blood smear image data to be recognized and a probability distribution map for characterizing the position distribution law of different types of blood cells on the blood smear;

[0050] An image recognition unit with a built-in recognition model, configured to be connected to the data storage unit for data connection, retrieve the blood smear image data, and recognize the blood smear image based on the recognition model, determine and label the cell types, accurate probabilities, and position coordinates corresponding to each cell image therein;

[0051] A reliability evaluation unit configured to be connected to the image recognition unit for data connection, used to obtain the cell types and accurate probabilities corresponding to each cell image after determination, as well as the position coordinates of each cell image, assign a reliability weight to the accurate probability in combination with the probability distribution map, and generate a type determination reliability value corresponding to each cell image;

[0052] A counting generation unit with a built-in comparison and output module, configured to be connected to the reliability evaluation unit for data connection, receive the type determination reliability value corresponding to each cell image and compare it with a set threshold: if it exceeds the set threshold, the corresponding cell image is determined as a definite sample and its cell type is labeled; if it does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the above sample to be determined are confirmed and labeled by manual and / or extended determination steps; based on the labeling information of each cell image in the blood smear image, count and output the quantity of each type of blood cell;

[0053] Among them, the blood smear corresponding to the blood smear image data is automatically made by a smearer, and the same blood smear making parameters are used when making the blood smear.

[0054] Through the above technical solution, a reliability weight value can be assigned to the determination result of the sample to be determined according to the position distribution law of different types of blood cells on the blood smear, thereby increasing the reliability of the determination result, making the quantity of each type of cell finally counted more reliable, and providing more reliable data support for doctors to diagnose the condition.

[0055] Further, the data storage unit further stores: a second correlation relationship among the overlapping probabilities of various types of cells, the overlapping positions of the cells, and the blood smear production parameters, and a first correlation relationship between the cell overlapping image and the combination of the cell types and quantities constituting the cell overlapping image;

[0056] The blood cell quantity counting system further includes:

[0057] An extended data preprocessing unit, configured to be connected to the data storage unit and the image recognition unit for data connection, obtain a cell overlapping image, obtain a combination of cell types and quantities corresponding to the cell overlapping image according to the first correlation relationship, and configure a first accurate probability value for each combination; obtain the position coordinates of the cell overlapping image, configure a second accurate probability value for each combination of cell types and quantities according to the second correlation relationship, and configure a third accurate probability value for each cell type in the above combination according to the probability distribution diagram; obtain the blood smear production parameter data corresponding to the current blood smear image, and configure a fourth accurate probability value for each combination of cell types and quantities according to the second correlation relationship;

[0058] A first extended determination unit, configured to be connected to the extended data preprocessing unit for data connection, assign reliable weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, and the fourth accurate probability value, calculate and obtain the cell type, quantity, and class quantity determination reliable value corresponding to the cell overlapping image according to a set algorithm; compare the class quantity determination reliable value with a set threshold, and determine the cell category and quantity corresponding to the cell overlapping image according to the comparison result, or determine the cell category and quantity corresponding to the cell overlapping image by manual determination.

[0059] Through the above technical solution, incorporating the blood smear production parameters that affect the cell distribution and overlapping probability on the blood smear into the determination probability analysis of uncertain samples can improve the reliability of the determination result.

[0060] Further, the blood cell quantity counting system further includes:

[0061] An image region segmentation unit, configured to divide all or part of the region of the blood smear image into a plurality of continuous and equal-area statistical regions;

[0062] A quantity proportion statistics unit, configured to be connected to the counting generation unit for data connection, and used to count the quantities of various types of cells included in the determined samples in each statistical region, and calculate the quantity proportion of each type of cell in the statistical region;

[0063] A reference curve generation unit configured to fit and generate a trend curve of the above ratio varying with the statistical region according to the ratio of the number of cells of a set type in each statistical region;

[0064] A disassembly and comparison unit configured to be data-connected to the extended data preprocessing unit, for obtaining the cell type and quantity combination corresponding to the cell overlap image in the statistical region, disassembling the cell type and quantity combination into the estimated quantity of each type of cell, adding the estimated quantity of each type of cell to the statistical data of the current statistical region, calculating the ratio of the quantity of each type of cell in the current statistical region before and after adding the estimated quantity, and configuring a fifth accuracy probability value for each cell type and quantity combination according to the deviation degree of the quantity ratio from the trend curve;

[0065] A second extension determination unit for assigning weight values to the first accuracy probability value, the second accuracy probability value, the third accuracy probability value, the fourth accuracy probability value, and the fifth accuracy probability value, and calculating and obtaining the cell type and quantity contained in the cell overlap image according to a set algorithm.

[0066] Through the above technical solution, the reliability of the blood cell quantity counting result can be further improved.

[0067] A blood cell quantity counting device based on image recognition, comprising:

[0068] An image acquisition device configured to acquire and output a blood smear image;

[0069] A blood smear production parameter acquisition device configured to be data-connected to a smearer for automatically producing a blood smear, for acquiring and outputting blood smear production parameter data;

[0070] The above-mentioned blood cell quantity counting system based on image recognition, configured to be data-connected to the image acquisition device and the blood smear production parameter acquisition device, for acquiring a blood smear image and blood smear production parameter data, and outputting the quantity and annotation information of each type of blood cell after being processed by a built-in program algorithm module; and

[0071] An information interaction device configured to be data-connected to the blood cell quantity counting system, for receiving a user's operation instruction and displaying the quantity and annotation information of each type of blood cell fed back from the system end.

[0072] A computer-readable storage medium on which a program module for implementing the above-mentioned blood cell quantity counting method based on image recognition is loaded.

[0073] Through the above technical solution, it is helpful to promote the use of the method of this application.

[0074] In summary, this application includes at least one of the following beneficial technical effects:

[0075] (1) Assign reliability weight values to the determination results of the sample to be determined according to the position distribution rules of different types of blood cells on the blood smear, thereby increasing the reliability of the determination results, making the number of each type of cell finally statistically obtained more reliable, and providing more reliable data support for doctors to diagnose the condition;

[0076] (2) Introduce the influencing factors that affect the cell overlap probability and the cell coincidence position into the determination of the cell type of the sample to be determined, effectively improving the determination accuracy of the cell type corresponding to the sample to be determined. Description of the Drawings

[0077] Figure 1 is a schematic diagram of the cell morphology on the blood smear;

[0078] Figure 2 is a schematic flowchart of the blood cell count method of the present application;

[0079] Figure 3 is a probability distribution diagram of the appearance probability of a set type of cells at different positions on the blood smear;

[0080] Figure 4 is a schematic diagram of the method for improving the reliability of blood cell count;

[0081] Figure 5 is a schematic diagram of the functional modules of the blood cell count system of the present application;

[0082] Figure 6 is a schematic diagram of the blood cell count device of the present application.

[0083] Reference numerals: 1, data storage unit; 2, image recognition unit; 3, reliability evaluation unit; 4, count generation unit; 5, extended data preprocessing unit; 6, first extended determination unit; 7, image region segmentation unit; 8, quantity ratio statistics unit; 9, reference curve generation unit; 10, disassembly comparison unit; 11, second extended determination unit; 100, image acquisition device; 200, blood smear production parameter acquisition device; 300, information interaction device; 400, blood cell count system. Detailed Embodiments

[0084] The following details the embodiments of the present application, and the examples of the embodiments are shown in the drawings.

[0085] In the description of this specification, the descriptions referring to terms such as "certain embodiments", "one embodiment", "some embodiments", "schematic embodiments", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the said embodiments or examples are included in at least one embodiment or example of this application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0086] Before elaborating on the embodiments of this application in detail, in order to more clearly illustrate the working principle of the embodiments of this application, the relevant technologies or basic theories related to blood cell morphology identification and counting are described herein.

[0087] There are three main types of cells in the blood: red blood cells, white blood cells, and platelets (thrombocytes). Red blood cells are the most numerous cells in the blood, and their main function is to transport oxygen and carbon dioxide between the lungs and various functional tissues of the body. White blood cells are an important part of the immune system and have multiple types, including neutrophils, lymphocytes, monocytes, and eosinophils, etc., and each type of cell has its corresponding morphology. Platelets are the smallest blood components and play a key role in the process of hemostasis and blood coagulation.

[0088] Currently, most disease diagnosis and treatment processes require complete blood cell analysis. The main method is to make a blood smear and then observe the quantity and morphology of various types of cells under a microscope, which is used as the basis for disease diagnosis.

[0089] The specific process of making a blood smear will not be elaborated here. Currently, the making of a blood smear can be automatically completed by a smearer. It should be noted here that various types of blood cells will show different distribution rules on the blood smear. For example, a large number of red blood cells are on the outermost layer of the blood smear, and there are some white blood cells and platelets in the middle area. White blood cells can be divided into granular and non-granular types according to their morphology. Among the granular white blood cells, neutrophils account for the largest proportion, followed by eosinophils and basophils. Among the non-granular white blood cells, monocytes have the largest volume, followed by lymphocytes, and platelets usually aggregate together to form small clusters. Different blood smear making parameters will form different cell distribution rules on the blood smear. The above blood smear making parameters mainly include: the volume of blood dropped on the glass slide, the angle between the spreader and the glass slide, and the moving speed of the spreader, etc. In an automatic smearer, precise control of the above parameters and even temperature, the dosage of the staining agent, and the pH value of the clear water can be achieved.

[0090] In the field of cell image recognition, by feeding a large number of cell images accurately labeled with cell type information into a deep learning algorithm, such as a convolutional neural network, a recognition model capable of accurately identifying various types of cells is trained. Currently, the recognition accuracy of the recognition model can reach over 95%. With the continuous progress of model training, the recognition accuracy of cell images will further increase in the future.

[0091] As Figure 1 shown in a blood smear image, it can be seen that the cell images are not completely independent and scattered. Many blood cells overlap, making it difficult to accurately determine the types and quantities of blood cells when using an image recognition model. As Figure 1 shown by the arrow in, the basophilic granulocytes are difficult to show the morphology and quantity of the overlapping cells behind them due to their darker staining. And Figure 1 the cells within the dashed box in have obvious partial overlap.

[0092] Based on the above introduction of related technologies and basic theories, the specific content of the embodiments of this application will be elaborated below.

[0093] The embodiments of this application disclose a method for counting the quantity of blood cells based on image recognition. As Figure 2 shown, it mainly includes the following steps:

[0094] S100, recognizing the blood smear image based on the recognition model, determining and labeling the cell types and accurate probabilities corresponding to each cell image therein;

[0095] S200, obtaining and generating a probability distribution map representing the probabilities of different types of blood cells appearing at different positions on the blood smear based on the position distribution rules of different types of blood cells on the blood smear;

[0096] S300, obtaining the cell types and accurate probabilities corresponding to each cell image after determination, as well as the position coordinates of each cell image, and assigning a reliability weight to the accurate probability in combination with the probability distribution map to generate a type determination reliability value corresponding to each cell image;

[0097] S400, comparing the type determination reliability value corresponding to each cell image with a set threshold:

[0098] S410, if the type determination reliability value corresponding to the cell image exceeds the set threshold, it is determined as a definite sample and its cell type is labeled;

[0099] S420, if the type determination reliability value corresponding to the cell image does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the above sample to be determined are confirmed and labeled by manual and / or extended determination steps;

[0100] S500, based on the annotation information of each cell image in the blood smear image, counts and outputs the quantities of various types of blood cells.

[0101] In step S100, the recognition model uses an AI image analysis model, which is built-in with a blood cell recognition model trained based on deep learning algorithms. The above-mentioned blood cell recognition model has been trained with a large amount of annotated blood cell image data, and can accurately recognize various types of blood cells and their subtypes such as red blood cells, white blood cells, and platelets. At the same time, it can also accurately detect morphological abnormalities of cells, such as uneven size, abnormal shape, nuclear-cytoplasmic changes, etc., and output the cell type corresponding to the cell image. In a specific implementation manner, information such as abnormal features corresponding to the cell image can also be output.

[0102] In the implementation manner of this application, the cell types corresponding to each cell image are determined and labeled as multiple, and each determined cell type corresponds to an accurate probability. For example, a cell image is determined to be neutrophils - 75%, eosinophils - 25%, basophils - 0%.

[0103] In a specific implementation manner, step S100, which recognizes the blood smear image based on the recognition model, further includes an image preprocessing step. Specifically, it includes operations such as noise reduction, contrast enhancement, and gray correction on the obtained blood smear image, removes the noise interference generated during the image acquisition process, highlights the morphological details of blood cells, makes the image features easier to extract, and then outputs the preprocessed blood smear image to the recognition model. When the to-be-recognized blood smear image obtained is not clear, the above steps can improve the recognition accuracy of the subsequent recognition model and the reliability of the counting result.

[0104] Combined Figure 3 As shown, for the same type of cell, the blood smear image is divided into multiple different probability regions, and the probabilities of various types of cells appearing at different positions on the blood smear can be obtained by theoretical analysis or big data statistics. Blood cells of different volumes and shapes will show different appearance probabilities at different positions on the glass slide when the blood is pushed at different moving speeds or tilting angles during the smear preparation.

[0105] In step S300, in order to ensure that the coordinate systems are unified during each blood cell morphological screening analysis, in practical applications, the blood sample for analysis is dropped at the same position on the glass slide, and the same blood smear preparation parameters are used when making the blood smear, such as the speed of the smear movement and the tilting angle. Then, taking the position where the blood sample is dropped, that is, the initial position, as the origin of the two-dimensional plane coordinate system, and the length and width directions of the glass slide as the x-axis and y-axis, the position coordinates corresponding to each cell image are counted.

[0106] Step S300, assign reliability weights to the accurate probabilities corresponding to each cell type in combination with the probability distribution map to generate a type determination reliability value for each cell image. In a simple implementation manner, it includes directly adding the accurate probability corresponding to each cell type to the occurrence probability at the position coordinate where it is located to obtain the type determination reliability value for each cell image. For example: for the A cell image, after image recognition, the probability of being determined as an eosinophil is 70%, and the probability of being determined as a neutrophil is 30%; while at the position where the A cell image is located, the occurrence probability of eosinophils is 20%, and the occurrence probability of neutrophils is 30%. Then the finally obtained type determination reliability value of the A cell image being determined as an eosinophil is 0.9, and the type determination reliability value of being determined as a neutrophil is 0.5.

[0107] In step S400, compare the type determination reliability value obtained in step S300 with a set threshold, such as 0.8, and finally determine whether the above cell image is a sample with a determined type according to the comparison result. In practical applications, the set thresholds corresponding to each type of cell are not the same, and the set thresholds corresponding to each type of cell can be set based on a large amount of determination data according to the accuracy rate of the determined samples.

[0108] If the cell category corresponding to the cell image cannot be determined through S410, usually it is confirmed by manual microscopy, or it can also be further determined by an extended determination step. Finally, the annotation information of each cell image in the blood smear image is used as the counting basis for counting.

[0109] When screening blood cell morphology, the main reason affecting the screening and counting accuracy rate is the cell overlapping image. As Figure 1 shown, in the blood smear image, some cells will overlap in the upper and lower parts or even completely cover. When using the recognition model to analyze the above cell overlapping image, there are often deviations. Therefore, it needs to be supplemented that: the cell images described in the implementation manner of this application include single cell images and cell overlapping images. The recognition model stores the association relationships between single cell images and cell types, as well as between cell overlapping images and the cell type and quantity combinations constituting the above cell overlapping images, and stores them in the first relational database.

[0110] In step S100, determining and annotating the cell type and accurate probability corresponding to each cell image includes:

[0111] Obtain single cell images and cell overlapping images in the blood smear image, search and match in the first relational database for cell images similar to the above single cell images and cell overlapping images, obtain their corresponding cell types or cell type and quantity combinations, and annotate accurate probabilities for each matching result according to the similarity degree between cell images.

[0112] In the embodiments of the present application, the single cell image determination outputs the cell type and its accurate probability, while the cell overlapping image corresponds to a combination of at least two cell types, and the number of each type of cell in the above combination may not be the same. For example, the cell composition corresponding to a cell overlapping image includes two red blood cells and one white blood cell, that is, the above cell overlapping image corresponds to a combination of multiple cells. Similarly, for the convenience of later reliability determination, the accurate probability is also marked for the above combination.

[0113] In the embodiments of the present application, in order to further improve the reliability of the blood cell count result, the blood cell quantity counting method further includes:

[0114] Establish the correlation relationship among the overlapping probabilities of each type of cell, the cell overlapping positions and the blood smear making parameters, and store it as the second relational database. The above correlation relationship can be obtained by training and learning big data through a deep neural network.

[0115] In step S420, the extension determination step includes:

[0116] S421, obtain the cell overlapping image, obtain multiple corresponding cell type and quantity combinations through the first relational database, and configure the first accurate probability value P1 for each combination. For example, for a cell overlapping image, according to the determination result of the first relational database, it can be: a combination of 1 eosinophil and 1 red blood cell, and the corresponding first accurate probability value P1 is 80%, or a combination of 1 eosinophil and 2 red blood cells, and the corresponding first accurate probability value P1 is 20%.

[0117] S422, obtain the position coordinates of the cell overlapping image, and configure the second accurate probability value P2 for each cell type and quantity combination according to the second relational database and the probability distribution map, and configure the third accurate probability value P3 for the cell types in the combination. The above second accurate probability value P2 is the probability value of the combination of cells corresponding to the cell overlapping image, while the third accurate probability value P3 is the probability value of a single type of cell in the combination. For example, according to the position coordinates of the cell overlapping image, the second accurate probability value P2 of its combination of 1 neutrophil and 1 platelet is 60%, while the probability of a single neutrophil appearing at the above position coordinates is 30%, and the probability of a platelet appearing at the above position coordinates is 90%.

[0118] S423, obtain the blood smear making parameter data corresponding to the current blood smear image, and configure the fourth accurate probability value P4 for each cell type and quantity combination according to the second relational database. Among them, the blood smear making parameters include the included angle between the spreader and the glass slide, the moving speed of the spreader, the environmental temperature in the instrument, and the blood volume, etc.

[0119] S424, Assign reliable weight values to the first accurate probability value P1, the second accurate probability value P2, the third accurate probability value P3, and the fourth accurate probability value P4, and calculate and obtain the cell type, quantity, and reliable value K of the class quantity determination corresponding to each cell overlapping image according to the set algorithm. In an embodiment, the above-mentioned reliable value K of the class quantity determination = 0.8P1 + 0.3P2 + 0.4P3 + 0.2P4. After calculating by the above formula, the reliable value K of the class quantity determination corresponding to each type of cell or its combination can be obtained, and then the cell type and quantity with the largest reliable value K of the class quantity determination are taken as the cell type and quantity corresponding to the cell overlapping image.

[0120] S425, Compare the above-mentioned reliable value K of the class quantity determination with a set threshold:

[0121] S4251, If the reliable value K of the class quantity determination corresponding to the cell overlapping image exceeds the set threshold, it is determined as a determined sample and its corresponding cell type and quantity are marked;

[0122] S4252, If the reliable value K of the class quantity determination corresponding to the cell overlapping image does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the above-mentioned sample to be determined are confirmed manually and marked.

[0123] The above technical solution introduces the influencing factors that affect the cell overlapping probability and the cell coincidence position into the determination of the cell type of the sample to be determined, which can effectively improve the determination accuracy of the cell type corresponding to the sample to be determined and provide reliable data support for the doctor's subsequent diagnosis.

[0124] In practical applications, the distribution of various types of blood cells on the blood smear is usually uniformly and continuously changing, that is, the ratio of the quantity of each type of cell in each statistical region should be a smooth curve. Based on the above distribution characteristics, in the embodiment of the present application, as Figure 4 shown, the method for counting the quantity of blood cells further includes:

[0125] S430, Divide all or part of the area of the blood smear image into a plurality of continuous and equal-area statistical regions. The above-mentioned statistical regions are preferably rectangular and arranged along the length direction of the blood smear. Among them, the standard for dividing the statistical regions is: the sum of the differences in the quantity of the samples to be determined in each adjacent statistical region is the largest.

[0126] S431, Count the quantity of each type of cell included in the determined samples in each statistical region, and calculate the proportion of the quantity of each type of cell in the statistical region. For example, the proportion of red blood cells is 60%, the proportion of white blood cells is 35%, and the proportion of platelets is 5%. Among them, the proportion of neutrophils in white blood cells is 50%, the proportion of lymphocytes is 20%, and the proportion of monocytes is 3%, etc.

[0127] S432. Fit and generate a trend curve of the above ratio changing with the statistical region based on the ratio of the number of cells of the set type in each statistical region. The horizontal axis of the above trend curve is the number information of the statistical region, and the vertical axis is the ratio of the number of cells of the set type.

[0128] S433. Obtain the cell type and quantity combination corresponding to the cell overlap image of the statistical region, and disassemble the above cell type and quantity combination into the estimated quantity of each type of cell. For example, disassemble a cell overlap image into 1 red blood cell and 1 neutrophil.

[0129] S434. Add the estimated quantity of each type of cell to the statistical data of the current statistical region, and calculate the ratio of the number of each type of cell in the current statistical region before and after adding the above estimated quantity.

[0130] S435. Configure a fifth accuracy probability value for each of the above cell type and quantity combinations according to the deviation degree between the above ratio of the number of cells and the trend curve. In practical applications, the above fifth accuracy probability value is inversely proportional to the deviation degree, that is, the greater the deviation degree caused by adding the disassembled cell type and quantity to the statistical region, the lower the reliability of the disassembled cell type and quantity.

[0131] S436. Assign weight values to the first accuracy probability value, the second accuracy probability value, the third accuracy probability value, the fourth accuracy probability value, and the fifth accuracy probability value, calculate and obtain the cell type, quantity, and class quantity determination reliability value in the cell overlap image according to the set algorithm, and finally determine the final cell type and quantity according to the class quantity determination reliability value. In the implementation manner of this application, the calculation method of step S436 is the same as that of step S424, and the weight values can be adjusted slightly according to the actual situation to make the final determination result more accurate, and the specific process will not be elaborated here.

[0132] Through the above technical solution, when the number of each type of cell inferred from the leukocyte overlap image is added to the existing number of each type of cell, if the ratio of the number of each type of cell changes too much, that is, the ratio of the number of cells in the current statistical region deviates significantly from the trend curve, it can be inferred that the number of each type of cell inferred by reverse inference does not satisfy the probability distribution law in statistics, so the reliability of the inference is greatly reduced, and vice versa, the reliability increases. At the same time, even if some cell images in the blood smear image are difficult to determine their original cell composition due to cell overlap for both manual or AI image recognition, it is possible to assign corresponding reliability weight values to various different cell type and quantity combinations through probability distribution, thereby further improving the reliability of cell type and quantity statistics.

[0133] To facilitate the rapid review of the quantity and morphology of blood cells of a set type by inspectors, the method for counting the number of cells described in this application further includes the following steps: receiving and responding to an external image review request, querying the target cell type based on the annotation information of each cell image in the blood smear image, and performing enhancement or fading processing on the set cell images. For example, if the inspector inputs the cell type that needs to be determined again, the system will then brighten and magnify the cells of the above type on the display screen, enabling the inspector to review the above cells faster and improving the efficiency of the entire blood cell counting.

[0134] To implement the above method for counting the number of blood cells based on image recognition, an embodiment of this application also discloses a system 400 for counting the number of blood cells based on image recognition, as Figure 5 shown, mainly including the following functional units: a data storage unit 1, an image recognition unit 2, a reliability evaluation unit 3, and a counting generation unit 4.

[0135] The data storage unit 1 is configured to obtain and store the blood smear image data to be recognized, as well as a probability distribution map for characterizing the position distribution law of different types of blood cells on the blood smear. In practical applications, the above data storage unit 1 is configured as an internal data storage hard disk of a PC or an external database. The above data storage unit 1 is data-connected to an external image acquisition device through a data interface, such as a high-precision microscope imaging device, and can quickly and clearly acquire the blood cell images of the blood smear, ensuring uniform illumination and accurate focus during the acquisition process to obtain high-quality original image data for subsequent analysis. In one embodiment, the blood cell counting system 400 described in this application may include the above image acquisition device. In another embodiment, the cell counting system described in this application may also be configured in a cloud server, directly receive the blood smear images uploaded by each terminal for analysis, and then send the analysis results to each terminal.

[0136] The image recognition unit 2 has a built-in recognition model and is configured to be data-connected to the data storage unit 1, retrieve the blood smear image data, and recognize the blood smear image based on the recognition model, determine and annotate the cell type, accurate probability, and position coordinates corresponding to each cell image therein. In practical applications, an image preprocessing module is also configured in the image recognition unit 2 to perform operations such as noise reduction, contrast enhancement, and gray correction on the acquired original image, remove the noise interference generated during the image acquisition process, highlight the morphological details of the blood cells, make the image features easier to extract, and the preprocessed image enters the recognition model for recognition. In a specific embodiment, the recognition model is configured in a cloud server, and the above image preprocessing process can be performed on each terminal, and then the preprocessed blood smear image is uploaded to the recognition model in the cloud server for recognition processing.

[0137] The reliability evaluation unit 3 is configured to be connected to the image recognition unit 2 for data connection, and is used to obtain the cell type, accurate probability corresponding to each cell image after determination, and the position coordinates of each cell image. Then, in combination with the probability distribution map, a reliability weight is assigned to the accurate probability to generate a type determination reliability value corresponding to each cell image.

[0138] The counting and generating unit 4 is built-in with a comparison and output module, which is configured to be connected to the reliability evaluation unit 3 for data connection, receive the type determination reliability value corresponding to each cell image, and compare it with a set threshold: if it exceeds the set threshold, the corresponding cell image is determined as a definite sample and its cell type is marked; if it does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the above sample to be determined are confirmed by manual and / or extended determination steps and marked. Finally, based on the annotation information of each cell image in the blood smear image, the quantity of each type of blood cell is counted and output. In practical applications, the above set threshold can be fine-tuned by the inspector according to the actual situation to make the final determination result more accurate.

[0139] In order to further improve the reliability of the recognition result, the data storage unit 1 also stores: the second correlation relationship among the overlapping probabilities of each type of cell, the cell overlapping positions, and the blood smear production parameters, and the first correlation relationship between the cell overlapping image and the combination of the cell types and quantities constituting the above cell overlapping image.

[0140] The blood cell quantity counting system 400 described in this application further includes an extended data preprocessing unit 5 and a first extended determination unit 6.

[0141] The extended data preprocessing unit 5 is configured to be connected to the data storage unit 1, the image recognition unit 2, and a smearer for automatically making blood smears externally for data connection, obtain cell overlapping images, and obtain multiple combinations of cell types and quantities corresponding to the above cell overlapping images according to the above first correlation relationship, and configure a first accurate probability value for each combination. At the same time, obtain the position coordinates of the cell overlapping images, configure a second accurate probability value for each combination of cell types and quantities according to the second correlation relationship, and then configure a third accurate probability value for each cell type in the combination according to the aforementioned probability distribution map. Finally, obtain the blood smear production parameter data corresponding to the current blood smear image from the smearer, such as the pushing speed and angle, etc., and configure a fourth accurate probability value for each combination of cell types and quantities according to the second correlation relationship. In a specific embodiment, the production of the blood smear is completed manually. For this reason, the system is also configured with a blood smear production parameter input unit for manually inputting the corresponding blood smear production parameters.

[0142] The first extension determination unit 6 is configured to be connected to the extension data preprocessing unit 5 for data connection, and assign reliable weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, and the fourth accurate probability value. Then, according to a set algorithm, calculate and obtain the cell type, quantity, and reliable determination value of the class quantity corresponding to the cell overlap image. Finally, compare the reliable determination value of the class quantity with a set threshold, and determine the cell category and quantity corresponding to the cell overlap image according to the comparison result, or determine the cell category and quantity corresponding to the cell overlap image by manual determination.

[0143] In the specific implementation process, both the above-mentioned extension data preprocessing unit 5 and the first extension determination unit 6 implement specific functions through specific computer program modules.

[0144] In the above technical solution, incorporating the blood smear production parameters that affect the cell distribution and overlap probability on the blood smear into the determination probability analysis of uncertain samples can significantly improve the reliability of the determination results.

[0145] Further optimized, the blood cell quantity counting system 400 further includes an image region segmentation unit 7, a quantity proportion statistics unit 8, a reference curve generation unit 9, a disassembly comparison unit 10, and a second extension determination unit 11.

[0146] The image region segmentation unit 7 is configured to divide all or part of the blood smear image into a plurality of continuous and equal-area statistical regions. Preferably, the above-mentioned statistical regions are rectangular and arranged along the length direction of the blood smear. The quantity proportion statistics unit 8 is configured to be connected to the counting generation unit 4 for data connection, and is used to count the quantity of each type of cell included in the determined samples in each statistical region, and calculate the quantity proportion of each type of cell in the statistical region. The reference curve generation unit 9 is configured to fit and generate a trend curve of the above ratio changing with the statistical region according to the ratio of the quantity proportion of the set type of cell in each statistical region. The disassembly comparison unit 10 is configured to be connected to the extension data preprocessing unit 5 for data connection, and is used to obtain the cell type and quantity combination corresponding to the cell overlap image in the statistical region, disassemble the above cell type and quantity combination into the estimated quantity of each type of cell, and then add the estimated quantity of each type of cell to the statistical data of the current statistical region, calculate the ratio of the quantity proportion of each type of cell in the current statistical region before and after adding the above estimated quantity, and configure a fifth accurate probability value for each cell type and quantity combination according to the deviation degree of the above quantity proportion ratio from the trend curve.

[0147] The second extension determination unit 11 is configured to assign weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, the fourth accurate probability value, and the fifth accurate probability value, and then calculate and obtain the cell types and quantities included in the cell overlap image according to a set algorithm. The specific implementation process has been described in the method embodiment and will not be elaborated here.

[0148] In an embodiment of the present application, an image recognition-based blood cell quantity counting device is also disclosed. As Figure 6 shown, the device mainly includes: an image acquisition device 100, a blood smear production parameter acquisition device 200, an information interaction device 300, and the aforementioned image recognition-based blood cell quantity counting system 400.

[0149] In detail, the image acquisition device 100 is configured to acquire and output a blood smear image. The aforementioned image acquisition device 100 can be configured as a high-precision microscope imaging device, or directly configured as an image data acquisition interface to share the blood smear image data to be recognized by other terminals.

[0150] The blood smear production parameter acquisition device 200 is configured to be data-connected to a smearer for automatically producing a blood smear, and is used to acquire and output blood smear production parameter data. In a specific embodiment, the aforementioned blood smear production parameter acquisition device 200 further includes a parameter manual input module configured in the information interaction device 300, and the inspection personnel manually input the production parameters of the blood smear.

[0151] The blood cell quantity counting system 400 is configured to be data-connected to the image acquisition device 100 and the blood smear production parameter acquisition device 200, and is used to acquire the blood smear image and the blood smear production parameter data, and output the quantities and annotation information of various types of blood cells after being processed by the built-in program algorithm module. In practical applications, the body of the aforementioned blood cell quantity counting system 400 can be configured in a cloud server, and directly configured as a data interaction module in a local blood cell quantity counting device. After verifying the user's authority, it receives the blood smear image data uploaded by the user and then feeds back the corresponding counting and analysis results according to the requirements.

[0152] The information interaction device 300 is configured to be data-connected to the blood cell quantity counting system 400, and is used to receive the user's operation instructions and display the quantities and annotation information of various types of blood cells fed back from the system end. In practical applications, the aforementioned information interaction device 300 can be configured as a PC or a tablet computer with network data communication functions.

[0153] For the convenience of popularizing and using the blood cell count method involved in the embodiments of the present application, a computer-readable storage medium is also proposed in the present application, on which a program module for implementing the blood cell count method based on image recognition as described above is loaded. In practical applications, the above computer-readable storage medium includes, but is not limited to, disk memories, CD-ROMs, optical memories, etc.

[0154] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A method for counting the number of blood cells based on image recognition, characterized in that, Including: Identifying a blood smear image based on an identification model, determining and labeling the cell type and accurate probability corresponding to each cell image therein; The production of the above blood smear is automatically completed by a smearer, and the same blood smear production parameters are used when making the blood smear; Obtaining and generating a probability distribution map for characterizing the appearance probabilities of different types of blood cells at different positions on the blood smear based on the position distribution rules of different types of blood cells on the blood smear; Obtaining the cell type and accurate probability corresponding to each cell image after determination, as well as the position coordinates of each cell image, and assigning a reliability weight to the accurate probability in combination with the probability distribution map to generate a type determination reliability value corresponding to each cell image; Comparing the type determination reliability value corresponding to each cell image with a set threshold: If the type determination reliability value corresponding to the cell image exceeds the set threshold, it is determined as a confirmed sample and its cell type is labeled; If the type determination reliability value corresponding to the cell image does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the above sample to be determined are confirmed and labeled by manual and / or extended determination steps; Based on the labeling information of each cell image in the blood smear image, counting and outputting the quantity of each type of blood cell; wherein, the cell image includes a single cell image and a cell overlap image; In the identification model, the association relationships between single cell images and cell types, and between cell overlap images and the combination of cell types and quantities constituting the above cell overlap images are stored as a first relational database; In the blood cell quantity counting method, the cell type corresponding to each cell image is determined and labeled as multiple, and each determined cell type corresponds to an accurate probability; Determining and labeling the cell type and accurate probability corresponding to each cell image includes: Obtaining single cell images and cell overlap images in the blood smear image, searching in the first relational database for matching cell images similar to the above single cell images and cell overlap images, obtaining the corresponding cell type or the combination of cell types and quantities, and labeling the accurate probability for each matching result according to the similarity degree between cell images.

2. The method for counting the number of blood cells based on image recognition according to claim 1, wherein The blood cell quantity counting method further includes: Establishing the association relationship among the overlap probabilities of each type of cell, the cell overlap positions, and the blood smear production parameters, and storing it as a second relational database; The extended determination step includes: Obtaining a cell overlap image, obtaining multiple corresponding combinations of cell types and quantities through the first relational database, and configuring a first accurate probability value for each combination; Obtaining the position coordinates of the cell overlap image, configuring a second accurate probability value for each combination of cell types and quantities according to the second relational database and the probability distribution map, and configuring a third accurate probability value for the cell types in the combination; Obtaining the blood smear production parameter data corresponding to the current blood smear image, and configuring a fourth accurate probability value for each combination of cell types and quantities according to the second relational database; Assign reliable weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, and the fourth accurate probability value, and calculate and obtain the cell type, quantity, and reliable value of class quantity determination corresponding to each cell overlapping image according to a set algorithm; Compare the reliable value of class quantity determination with a set threshold: If the reliable value of class quantity determination corresponding to the cell overlapping image exceeds the set threshold, it is determined as a definite sample and its corresponding cell type and quantity are labeled; If the reliable value of class quantity determination corresponding to the cell overlapping image does not exceed the set threshold, it is determined as a sample to be determined, and the cell type and quantity of the sample to be determined are manually confirmed and labeled; Among them, the blood smear making parameters include the included angle between the spreader and the glass slide, the moving speed of the spreader, and the blood volume.

3. The method for counting the number of blood cells based on image recognition according to claim 2, wherein The method for counting the number of blood cells further includes: Divide all or part of the area of the blood smear image into multiple continuous and equal-area statistical regions; Count the number of each type of cell contained in the definite samples in each statistical region, and calculate the proportion of the number of each type of cell in the statistical region; According to the ratio of the proportion of the set type of cell in each statistical region, fit and generate a trend curve of the ratio changing with the statistical region; Obtain the cell type and quantity combination corresponding to the cell overlapping image in the statistical region, and disassemble the cell type and quantity combination into the estimated quantity of each type of cell; Add the estimated quantity of each type of cell to the statistical data of the current statistical region, and calculate the ratio of the proportion of the number of each type of cell in the current statistical region before and after adding the estimated quantity; According to the degree of deviation between the ratio of the proportion of the number and the trend curve, configure a fifth accurate probability value for each cell type and quantity combination; Assign weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, the fourth accurate probability value, and the fifth accurate probability value, and calculate and obtain the cell type and quantity contained in the cell overlapping image according to a set algorithm; Among them, the standard for dividing statistical regions is: the sum of the differences in the number of samples to be determined in each adjacent statistical region is the largest.

4. The method for counting the number of blood cells based on image recognition according to claim 1, wherein The method for counting the number of cells further includes: Receive and respond to an external image review request, and query the target cell type and perform enhancement or fading processing on the set cell image based on the annotation information of each cell image in the blood smear image.

5. A blood cell count system based on image recognition, characterized in that, Include: A data storage unit (1), configured to obtain and store the blood smear image data to be recognized, and a probability distribution map for characterizing the position distribution law of different types of blood cells on the blood smear; An image recognition unit (2), with a built-in recognition model, configured to be data-connected to the data storage unit (1), retrieve the blood smear image data, and recognize the blood smear image based on the recognition model, and determine and label the cell type, accurate probability, and position coordinates corresponding to each cell image therein; A reliability evaluation unit (3), configured to be data-connected to the image recognition unit (2), used to obtain the cell type and accurate probability corresponding to each cell image after determination, and the position coordinates of each cell image, and combine the probability distribution map to assign a reliability weight to the accurate probability, and generate a reliable value of type determination corresponding to each cell image; A counting generation unit (4) with a built-in comparison output module configured to be connected to the reliability evaluation unit (3) for data connection, receive the type determination reliability value corresponding to each cell image, and compare it with a set threshold: if it exceeds the set threshold, determine the corresponding cell image as a determined sample and label its cell type; if it does not exceed the set threshold, determine it as a sample to be determined, and confirm the cell type and quantity of the above sample to be determined by manual and / or extended determination steps and label them; based on the labeling information of each cell image in the blood smear image, count and output the quantity of each type of blood cell; wherein, the blood smear corresponding to the blood smear image data is automatically made by a smearer, and the same blood smear making parameters are used when making the blood smear; The data storage unit (1) also stores: the second correlation relationship among the overlapping probabilities of each type of cell, the cell overlapping positions, and the blood smear making parameters, and the first correlation relationship between the cell overlapping image and the combination of the cell type and quantity constituting the above cell overlapping image; The blood cell quantity counting system (400) further includes: An extended data preprocessing unit (5) configured to be connected to the data storage unit (1) and the image recognition unit (2) for data connection, obtain the cell overlapping image, and obtain multiple combinations of cell types and quantities corresponding to the above cell overlapping image according to the above first correlation relationship, and configure a first accurate probability value for each combination; obtain the position coordinates of the cell overlapping image, configure a second accurate probability value for each combination of the cell types and quantities according to the second correlation relationship, and configure a third accurate probability value for each cell type in the above combination according to the probability distribution diagram; obtain the blood smear making parameter data corresponding to the current blood smear image, and configure a fourth accurate probability value for each combination of the cell types and quantities according to the second correlation relationship; A first extended determination unit (6) configured to be connected to the extended data preprocessing unit (5) for data connection, assign reliable weight values to the first accurate probability value, the second accurate probability value, the third accurate probability value, and the fourth accurate probability value, calculate and obtain the cell type, quantity, and class quantity determination reliability value corresponding to the cell overlapping image according to a set algorithm; compare the above class quantity determination reliability value with a set threshold, and determine the cell category and quantity corresponding to the above cell overlapping image according to the comparison result, or determine the cell category and quantity corresponding to the above cell overlapping image by manual determination.

6. The blood cell count system based on image recognition according to claim 5, wherein The blood cell quantity counting system further includes: An image region segmentation unit (7) configured to divide all or part of the blood smear image into multiple continuous and equal-area statistical regions; A quantity proportion statistics unit (8) configured to be connected to the counting generation unit (4) for data connection, used to count the quantity of each type of cell included in the determined samples in each statistical region, and calculate the quantity proportion of each type of cell in the statistical region; A reference curve generation unit (9) configured to fit and generate a trend curve of the above ratio changing with the statistical region according to the ratio of the quantity proportion of the set type of cell in each statistical region; The disassembling and comparison unit (10) is configured to be data-connected to the extended data preprocessing unit (5), and is used to obtain the cell type and quantity combination corresponding to the cell overlapping image in the statistical area, disassemble the cell type and quantity combination into the estimated quantity of each type of cell, add the estimated quantity of each type of cell to the statistical data of the current statistical area, calculate the ratio of the quantity of each type of cell in the current statistical area before and after adding the estimated quantity, and configure the fifth accuracy probability value for each cell type and quantity combination according to the deviation degree of the quantity ratio from the trend curve; The second extended determination unit (11) is used to assign weight values to the first accuracy probability value, the second accuracy probability value, the third accuracy probability value, the fourth accuracy probability value and the fifth accuracy probability value, and calculate and obtain the cell type and quantity contained in the cell overlapping image according to the set algorithm.

7. A blood cell count device based on image recognition, characterized in that, Including: The image acquisition device (100) is configured to acquire and output a blood smear image; The blood smear production parameter acquisition device (200) is configured to be data-connected to the smearer for automatically producing a blood smear, and is used to acquire and output blood smear production parameter data; The image recognition-based blood cell quantity counting system (400) according to any one of claims 5-6 is configured to be data-connected to the image acquisition device (100) and the blood smear production parameter acquisition device (200), and is used to acquire a blood smear image and blood smear production parameter data, and output the quantity and annotation information of each type of blood cell after being processed by the built-in program algorithm module; And The information interaction device (300) is configured to be data-connected to the blood cell quantity counting system (400), and is used to receive the user's operation instruction and display and output the quantity and annotation information of each type of blood cell fed back from the system end.

8. A computer-readable storage medium, characterized in that, A program module for implementing the image recognition-based blood cell quantity counting method according to any one of claims 1-4 is loaded thereon.

Images