Blood analysis system

Abstract

The application provides a blood analysis system, comprising a sampling device, a sample preparation device, a detection device and a data processing device. The sampling device collects a blood sample and delivers the blood sample to the sample preparation device, the sample preparation device prepares a to-be-tested sample liquid, the detection device detects the to-be-tested sample liquid to obtain pulse signals of each cell, and the data processing device comprises a processor and a non-transitory computer readable storage medium storing a computer program. When the computer program is called by the processor, the following steps are performed: converting the pulse signals into blood cell detection data, inputting the blood cell detection data into a pre-trained deep learning neural network model, and performing classification, counting, correction, sample type detection or abnormality prompting and other processing operations by the deep learning neural network model.

Description

Technical Field

[0001] This application relates to the field of medical device technology, and more specifically, to blood analysis systems. Background Technology

[0002] A blood cell analyzer is a commonly used medical device that can be used to detect the number, type, and state of cells in the blood of a target individual. This information can serve as a diagnostic indicator of the target individual's health status.

[0003] Current blood cell analyzers typically employ three main detection methods: impedance analysis, laser scattering analysis, and chemical staining analysis. Impedance analysis uses the Coulter principle to acquire pulse signals corresponding to cell volume, which are then used for cell counting and classification. Laser scattering analysis uses a helium-neon laser source to scan each cell with a monochromatic laser, collecting lateral and forward scattered light signals. Forward scattered light can detect cell volume information, while lateral scattered light can detect the lobulation of the cell nucleus and the granulation of the cytoplasm, providing image information about cell granules. Chemical staining analysis involves staining cells with relevant chemical reagents; different cell types react differently, and the cell type is determined by measuring the absorption rate of the scattered light generated by laser irradiation.

[0004] The accuracy of the results obtained by these traditional digital signal processing methods or traditional image processing methods needs to be further improved. Summary of the Invention

[0005] In view of this, this application provides a blood analysis system and related methods to improve the accuracy of blood cell classification. To achieve this objective, this application provides the following technical solutions:

[0006] In a first aspect, embodiments of this application provide a blood analysis system, including:

[0007] A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device, wherein the blood samples include at least white blood cells and red blood cells;

[0008] A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested;

[0009] A detection device is used to detect the sample solution to obtain the pulse signal of the cells in the sample solution;

[0010] A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: obtaining blood cell distribution data based on the pulse signals of the cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain a classification result output by the deep learning neural network model.

[0011] Secondly, embodiments of this application provide a blood analysis system, including:

[0012] A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device;

[0013] A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested;

[0014] A detection device is used to detect the sample solution to obtain the pulse signal of the cells in the sample solution;

[0015] A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: generating blood cell distribution data based on the pulse signals of the cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain corrected blood cell distribution data output by the deep learning neural network model.

[0016] Thirdly, embodiments of this application provide a blood analysis system, including:

[0017] A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device;

[0018] A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested;

[0019] A detection device is used to detect the sample solution to obtain the pulse signal of the cells in the sample solution;

[0020] A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: generating blood cell detection data based on the pulse signals of the cells, and inputting the blood cell detection data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample.

[0021] Fourthly, embodiments of this application provide a blood analysis system, including:

[0022] A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device;

[0023] A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested;

[0024] A detection device is used to detect the sample solution to obtain the pulse signal of the cells in the sample solution;

[0025] A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: generating blood cell distribution data based on the pulse signals of the cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain a prompt indicating whether the measurement process of the blood sample is abnormal.

[0026] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0027] This application provides a blood analysis system that can use a deep learning neural network model to analyze and process blood cell detection data from blood samples. Since the deep learning neural network model is trained from training samples of blood cell detection data, the trained deep learning neural network model has the ability to analyze and process input blood cell detection data of the same type as the training samples. Furthermore, the deep learning neural network model uses a deep learning algorithm in its analysis and processing process, which can mine richer feature information from the input data, resulting in higher accuracy of the output analysis and processing results. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of a blood analysis system embodiment;

[0030] Figure 2 This is a schematic diagram of the data processing device in a blood analysis system;

[0031] Figure 3A Example image of a two-dimensional scatter plot obtained by combining three different optical signals;

[0032] Figure 3B A schematic diagram illustrating how a deep learning neural network model classifies a two-dimensional scattered image.

[0033] Figure 3C This is a schematic diagram of the structure of a deep learning neural network model;

[0034] Figure 4 Another schematic diagram illustrating the classification of two-dimensional scatter images by a deep learning neural network model;

[0035] Figure 5 Another schematic diagram illustrating the classification of two-dimensional scattered images by a deep learning neural network model;

[0036] Figure 6A A schematic diagram illustrating how a deep learning neural network model classifies a one-dimensional histogram.

[0037] Figure 6B This is another structural diagram of a deep learning neural network model;

[0038] Figure 7 Another schematic diagram illustrating the classification of one-dimensional histograms by a deep learning neural network model;

[0039] Figure 8 A schematic diagram illustrating the calculation of hemoglobin concentration using a deep learning neural network model;

[0040] Figure 9 Another schematic diagram illustrating the classification of two-dimensional scattered images by a deep learning neural network model;

[0041] Figure 10A A schematic diagram showing the platelet identification results for normal and abnormal samples;

[0042] Figure 10BA schematic diagram illustrating interference removal in a one-dimensional histogram of platelets;

[0043] Figure 10C This is a schematic diagram comparing the platelet one-dimensional histogram before and after interference removal.

[0044] Figure 11 A schematic diagram illustrating the correction of a two-dimensional scatter image;

[0045] Figure 12 This is another structural diagram of a deep learning neural network model;

[0046] Figure 13A A schematic diagram for selecting a two-dimensional scatter image from a specific viewpoint in a three-dimensional scatter image;

[0047] Figure 13B A schematic diagram illustrating two different detection results: the original alarm and the normal result;

[0048] Figure 13C Another schematic diagram to obtain two different detection results: the original alarm and the normal result;

[0049] Figure 14A Two-dimensional scatter images obtained from the DIFF channel, RET channel, and WNB channel respectively;

[0050] Figure 14B A schematic diagram showing that the DIFF channel can produce two different test results: one requiring re-inspection and one normal result.

[0051] Figure 14C A schematic diagram illustrating two different test results obtained by the RET channel: one requiring retesting and one normal result.

[0052] Figure 14D This is a schematic diagram showing two different test results obtained by the WNB channel: one requiring re-inspection and one normal result.

[0053] Figure 15 A schematic diagram illustrating the detection results for three different types of substances;

[0054] Figure 16 A schematic diagram illustrating two different test results: a blocked hole and a normal hole.

[0055] Figure 17 This is a schematic diagram to obtain two different test results: abnormal and normal sampling. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0057] This application provides a blood analysis system (also known as a blood cell analyzer). This system utilizes deep learning-based artificial intelligence technology to perform one or more processing operations on blood cell pulse signals, including but not limited to classification, counting, image transformation processing, and alarm functions. The structure of the blood analysis system performing different blood processing operations is described below according to different application scenarios.

[0058] To facilitate understanding of the technical solution, the Chinese meanings of the English concepts that may be involved in the embodiments of this application are first introduced in Table 1 below.

[0059] Table 1

[0060]

[0061] Application scenario 1: Blood cell classification.

[0062] See Figure 1 This illustrates a specific structure of a blood analysis system capable of classifying blood cells. The blood analysis system includes at least: a sampling device 110, a sample preparation device 120, a detection device 130, and a data processing device 140.

[0063] Sampling device 110 is used to collect blood samples and transport them to a sample preparation device. Specifically, the blood sample can be a human blood sample or a blood sample from other animals. The blood sample includes at least white blood cells and red blood cells. In one implementation, sampling device 110 may have a pipette (e.g., a sampling needle) with a pipette nozzle and a drive unit for driving the pipette to quantitatively draw the blood sample to be tested through the pipette nozzle. For example, the sampling needle may move to a sample container containing the blood sample under the drive unit and draw the blood sample to be tested from it.

[0064] The sample preparation apparatus 120 includes at least one reaction chamber and a reagent supply unit (not shown), wherein the reagent supply unit provides processing reagents, and the at least one reaction chamber provides a reaction site for the blood sample and the processing reagents to react and prepare a test sample solution. Specifically, at least one reaction chamber receives a blood sample collected by the sampling device 110, and the reagent supply unit provides processing reagents to the at least one reaction chamber, thereby mixing the blood sample with the processing reagents provided by the reagent supply unit in the reaction chamber to prepare a test sample solution.

[0065] It should be noted that the reagent supply unit may include one or more. To avoid reagent contamination, different reagent supply units can provide different processing reagents. Different processing reagents are mixed with blood samples for different purposes to prepare different test sample solutions. If some processing reagents are the same, then the reagent supply unit corresponding to the processing reagents can be the same reagent supply unit.

[0066] The detection device 130 is used to detect the sample solution to obtain the pulse signals of the cells in the sample solution. Specifically, after the sample preparation device 120 prepares the sample solution to be tested, the detection device 130 detects the sample solution to obtain the pulse signals of each cell in the sample solution.

[0067] In one specific implementation, the detection device includes an optical detection device capable of acquiring at least two types of optical pulse signals from cells in the sample solution. Specifically, the optical detection device includes an optical flow chamber, a light source, and a detector. Cells in the sample solution pass through the optical flow chamber one by one, and the detector collects the pulse signals emitted by the cells when illuminated by the light source. In another specific implementation, the detection device includes an impedance detection device capable of detecting changes in the electrical impedance of cells in the sample solution as they pass through a detection aperture, thereby acquiring pulse signals. The pulse signals detected by the impedance detection device are one-dimensional data. In yet another specific implementation, the detection device includes a hemoglobin detection device capable of detecting changes in the transmittance of the sample solution. It should be noted that the above detection devices are merely illustrative examples. Other detection devices that can detect cellular particles (which may be referred to as blood cells, cells, or particles), such as colorimetric detection devices, which are conceivable to those skilled in the art, are within the scope of protection of the embodiments of this application.

[0068] The data processing device 140 is connected to the detection device and includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: generating blood cell distribution data based on cell pulse signals, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain the classification result output by the deep learning neural network model.

[0069] Specifically, the data processing device 140 includes at least a processing component with data processing capabilities, such as a processor, and a computer-readable storage medium. The computer-readable storage medium stores a computer program and may store data related to the computer program, such as a pre-trained deep learning neural network model. When the computer program is invoked by the processor, the deep learning neural network model can be used to classify cells in a blood cell distribution map generated by cell pulse signals.

[0070] In one implementation, such as Figure 2As shown, the data processing device 140 includes at least a processing component 141, a random access memory (RAM) 142, a read-only memory (ROM) 143, a communication interface 144, an I / O interface 145, and a memory 146. The processing component 141, RAM 142, ROM 143, communication interface 144, I / O interface 145, and memory 146 communicate via a bus 147. The processing component can be a central processing unit (CPU), a graphics processing unit (GPU), or other chips with computing capabilities. The memory 146 stores various computer programs executed by the processor component 141 and the data required to execute those programs. Additionally, during blood sample analysis, data requiring local storage, such as pre-trained deep learning neural network models, can be stored in the memory 146. The I / O interface 145 consists of serial interfaces such as USB, IEEE 1394, or RS-232C; parallel interfaces such as SCSI, IDE, or IEEE 1284; and analog signal interfaces composed of D / A converters and A / D converters. The I / O interface 145 can connect to display devices with display functions, such as LCD screens, touch screens, and LED displays. The data processing device 140 can output obtained data, such as blood cell distribution images (one-dimensional histograms, two-dimensional scatter plots, and three-dimensional scatter plots), and classification results output by deep learning neural network models (such as cell type counts and cell type classification labels), to the display device for display. The communication interface 144 can be any known communication protocol, and it communicates with the outside world via a network. The data processing device 140 can transmit data with any device connected through the network via the communication interface 144 using a certain communication protocol. For example, the data processing device 140 can communicate with a computer device via the communication interface 144, and the computer device can send the trained deep learning neural network model to the data processing device 140 for storage; or the data processing device 140 can connect to a display via the communication interface 144, and send the processed data, such as a blood cell distribution image and the classification results output by the deep learning neural network model, to the display for output; or the data processing device 140 can communicate with a printer via the communication interface 144, and send the processed data, such as a blood cell distribution image and the classification results output by the deep learning neural network model, to the printer for printing.

[0071] After receiving the pulse signal collected by the detection device 130, the data processing device 140 first needs to obtain blood cell distribution data based on the pulse signal of the cells. The blood cell distribution data can be one-dimensional histogram data, two-dimensional scatter plot data, or three-dimensional scatter plot data, voltage curve data such as hemoglobin voltage curve data, etc.; alternatively, the blood cell distribution data can be blood cell distribution images such as histogram images or scatter plot images, or it can be a data array directly obtained based on the pulse signal, which can be further processed to obtain an image. The blood cell distribution data is used to describe the characteristics of blood cells from one or more dimensions, such as volume dimension and cell particle content dimension.

[0072] Specifically, if the detection device 130 is an optical detection device, after the optical detection device obtains at least two types of optical pulse signals from the cells, the data processing device 140 processes the pulse signals by obtaining blood cell scatter distribution data based on the pulse signals. The blood cell scatter distribution data can be blood cell scatter plot data or blood cell scatter distribution image, wherein the blood cell scatter distribution image is obtained by further processing the blood cell scatter plot data. Optionally, the optical pulse signals obtained by the optical detection device can be optical pulse signals in two dimensions: forward scattered light and side scattered light, or optical pulse signals in three dimensions: forward scattered light, side scattered light, and fluorescence intensity.

[0073] If the detection device 130 is an impedance detection device, after the impedance detection device obtains the pulse signal from the cell, the data processing device 140 obtains the blood cell impedance data based on the pulse signal. Similarly, the blood cell impedance data can be an impedance data array or a histogram. The histogram is a one-dimensional sequence obtained after further processing of the impedance data array.

[0074] If the detection device 130 is a hemoglobin detection device, after the hemoglobin detection device obtains the change in transmittance, the data processing device 140 obtains the hemoglobin voltage curve based on the change in transmittance.

[0075] After obtaining the blood cell distribution data, the data processing device 140 further inputs the blood cell distribution data into a pre-trained deep learning neural network model. The deep learning neural network model processes the blood cell distribution data based on a deep learning algorithm and outputs the classification results of the blood cells. The classification results may include one of the following: the correspondence between blood cell types and the cell particles included in the blood cell distribution data, and / or, the statistical results of blood cell types. The first scenario is outputting the blood cell type of the cell particles, such as outputting the classification label of the cell particles; the second scenario is outputting the statistical results of each blood cell type. It should be noted that the classification results are obtained by the deep learning neural network model through analysis and processing of the input blood cell distribution data, and are output by the deep learning neural network model. The analysis and processing at least includes the classification of the cell types included in the blood cell distribution data by the deep learning neural network model. In other words, this application can fully utilize the advantages of deep learning neural network models in data analysis and processing; the classification results can be directly obtained from the deep learning neural network model without further analysis and processing, making the method of obtaining classification results simpler and more direct.

[0076] It should be noted that, in this embodiment, a deep learning neural network model is pre-trained using training samples of blood cell distribution data (training samples include blood cell distribution data samples and classification result labels of the blood cell distribution data samples). The trained deep learning neural network model has the ability to classify different types of cell particles in the blood cell distribution data. The reason why the trained deep learning neural network model has this ability is that it can determine what data features to use to accurately distinguish different types of cell particles. In a specific embodiment, the data samples of the deep learning neural network model need to correspond to the data type input when applying the specific embodiment. In addition, the labels of the training samples determine the processing capabilities of the deep learning neural network model, such as blood cell classification, blood cell data correction, identification of blood sample type, and judgment of whether the measurement process is abnormal.

[0077] For example, assuming the blood cell distribution data is blood cell image data, a deep learning neural network model can extract image features to obtain a classification result. Image features may include any one or more of the following: particle shape, color, texture, grayscale, degree of aggregation, etc. For ease of description, the determined image features can be called target image features. Therefore, after obtaining the blood cell distribution image of the sample fluid to be tested, the blood cell distribution image is input into a pre-trained deep learning neural network model. The model can then automatically extract target image features from the blood cell distribution image, and then use these target image features to identify the classification result of the blood cell distribution image.

[0078] To provide users with a clear and intuitive understanding of the processing results, the blood analysis system may also include a display device mounted on the outer surface of the system's casing. Data processed by the data processing unit 140, such as blood cell distribution images, cell particle classification labels, and statistical results of cell types, can all be output and displayed via the display device.

[0079] Current blood cell analyzers typically use impedance spectroscopy and laser scattering to classify blood cells. Specifically, each blood cell is passed through a sensor-equipped detection aperture to obtain the impedance or optical signal of each cell at each sensor. Impedance spectroscopy generally obtains a one-dimensional signal, while laser scattering obtains a multi-dimensional signal corresponding to multiple directions. Based on the different pulse signals of different cell types in each dimension, blood cell counting and classification are achieved. However, the blood analysis system provided in this application includes a sampling device, a sample preparation device, a detection device, and a data processing device. The sampling device, sample preparation device, and detection device prepare the collected blood sample into a test sample solution, detect the cells in the test sample solution to obtain cell pulse signals, and the data processing device processes the pulse signals into blood cell distribution data. Then, a pre-trained deep learning neural network model processes the blood cell distribution data and outputs classification results. The deep learning neural network model, trained based on deep learning algorithms, has good cell classification capabilities, and its output classification results have high accuracy.

[0080] The blood analysis system will be described in detail below through several different specific embodiments.

[0081] The first type of blood cell analysis system can output the classification labels corresponding to particles.

[0082] In one implementation, the detection device 130 of the blood cell analysis system is an optical detection device, and the data processing device 140 obtains blood cell scatter distribution data based on the optical pulse signal collected by the optical detection device, and inputs the blood cell scatter distribution data into a deep learning neural network model to obtain the classification label corresponding to the cell particles in the blood cell scatter distribution data. The classification label is used to represent the cell type of the cell particles.

[0083] It should be noted that if the blood cell scatter distribution data is in the form of a blood cell scatter plot, the deep learning neural network model adds a corresponding classification label to each blood cell scatter plot data point. This classification label is used to indicate the cell type of the blood cell particles corresponding to that blood cell scatter plot data point. If the blood cell scatter distribution data is in the form of a blood cell scatter distribution image, the deep learning neural network model adds a corresponding classification label to the particle clusters in the blood cell scatter distribution image. This classification label is used to indicate the cell type of the blood cell particle clusters. For example, in a blood cell scatter distribution image, at least one particle cluster is marked with an outline, and the blood cell type of each particle cluster is output.

[0084] Specific Implementation Examples 1-1

[0085] The blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. These devices can form a DIFF detection channel for differentiating white blood cells in a blood sample into five categories: neutrophils, eosinophils, basophils, lymphocytes, and monocytes. Each blood cell in the sample passes through the optical detection device, obtaining optical signals from the blood cells in multiple dimensions, such as forward scattered light (FS), side scattered light (SS), and side fluorescence scattered light (FL). These optical signals can be specifically represented as optical pulse signals.

[0086] The data processing device converts the optical pulse signals of cells into a scatter distribution image of blood cells, which can also be simply referred to as a scatter distribution image. For example, after the optical detection channel detects optical pulse signals in two or more directional dimensions, the data processing device can combine the optical pulse signals from any two of the multiple directional dimensions into a two-dimensional scatter distribution image. For instance, lateral fluorescence scattering (FL) and lateral scattered light (SS) are combined into an FL-SS image, forward scattered light (FS) and lateral scattered light (SS) are combined into an FS-SS image, and lateral fluorescence scattering (FL) and forward scattered light (FS) are combined into an FL-FS image. See [link to documentation]. Figure 3A It provides examples of two-dimensional scatter distribution images obtained by the above three combination methods. The three images from left to right are FL-SS image, FS-SS image and FL-FS image respectively.

[0087] To improve the accuracy of deep learning neural network models in classifying cells in scatter plot images of blood cells, a data processing device can preprocess the blood cell scatter plot image before inputting it into the deep learning neural network model. This preprocessing includes, but is not limited to, filtering, denoising, and enhancement. The data processing device then inputs the preprocessed blood cell scatter plot image into the trained deep learning neural network model. The deep learning neural network model classifies the particle clusters in the blood cell scatter plot image and outputs the cell type labels corresponding to each particle cluster, such as neutrophils, eosinophils, basophils, lymphocytes, and monocytes. Figure 3B As shown, a scatter plot image of blood cells, composed of lateral fluorescence and lateral scattered light signals, is input into a deep learning neural network model. After downsampling and upsampling, the model outputs cell classification results from the scatter plot image, indicating the particle clusters corresponding to five cell types: neutrophils (Neu), eosinophils (Eos), basophils (Baso), lymphocytes (Lym), and monocytes (Mon). It should be noted that the training samples for the deep learning neural network model are scatter plot images of blood cells, and these samples include classification labels for neutrophils, eosinophils, basophils, lymphocytes, and monocytes.

[0088] For example, deep learning neural network models can specifically include convolutional neural networks such as SegNet, Unet, or DeepLab. See [link to documentation]. Figure 3C Taking the Unet network model as an example, the structure and classification steps of this deep learning neural network model are explained. The Unet network model is an encoder-decoder structure: the encoder structure consists of one or more downsampling layers, mainly used to extract image features such as grayscale, color, texture, and shape of each pixel position in the input blood cell scatter distribution image; the decoder structure consists of one or more upsampling layers, mainly used to recover the blood cell scatter distribution image based on the image features, and finally outputs the cell particle type corresponding to each pixel position. This deep learning neural network model can be applied to multiple embodiments of this application to identify various different types of cells included in the blood cell scatter distribution image. In this embodiment, the model identifies at least one of the five categories of white blood cells.

[0089] It should be noted that the data processing device may not input a direct image of blood cell scatter distribution into the deep learning neural network model, but rather blood cell scatter plot data that generates the blood cell scatter distribution image. The deep learning neural network model then adds cell type labels to the blood cell scatter plot data.

[0090] Specific embodiments 1-2

[0091] The blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. These devices can form a WNB detection channel for classifying basophils and white blood cells, or for classifying basophils, white blood cells, and optionally nucleated and nucleated red blood cells.

[0092] Similar to Specific Embodiment 1-1, the optical detection device can obtain optical pulse signals of at least two types of blood cells in the sample fluid to be tested. The data processing device can convert the optical pulse signals of at least two types of blood cells into a blood cell scatter distribution image, and input the blood cell scatter distribution image into a deep learning neural network model after preprocessing. Unlike Specific Embodiment 1-1, the blood cell scatter distribution image output by the deep learning neural network model includes basophils, white blood cells, and optionally nucleated red blood cells. Furthermore, the training samples used to train the deep learning neural network model in this specific embodiment are also different from those in Specific Embodiment 1-1; the training samples in this specific embodiment are blood cell scatter distribution images including basophils, white blood cells, and optionally nucleated red blood cells.

[0093] like Figure 4 As shown, a scatter image of blood cell distribution, composed of forward-scattered light signals and lateral fluorescence signals, is input into a deep learning neural network model. After downsampling and upsampling, the deep learning neural network model outputs the cell classification results of the scatter image of blood cells, specifically labeling the particle clusters corresponding to ghost cells (Ghost), white blood cells (Wbc), nucleated red blood cells (Nrbc), and basophils (Baso). It should be noted that in one embodiment, the labeling results of ghost cells may not be displayed. The deep learning neural network model used in this embodiment can have the same structure as that used in specific embodiment 1-1; for a description of this model, please refer to the above description, which will not be repeated here.

[0094] Specific embodiments 1-3

[0095] A blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. These devices can form a RET detection channel for classifying platelets, red blood cells, and reticulocytes.

[0096] Similar to specific embodiment 1-1, the optical detection device can obtain at least two types of optical pulse signals of blood cells in the sample fluid to be tested. The data processing device can convert the at least two types of optical pulse signals into a blood cell scatter distribution image, and input the blood cell scatter distribution image into a deep learning neural network model after preprocessing. Unlike specific embodiment 1-1, as... Figure 5As shown, the blood cell scatter distribution image output by the deep learning neural network model includes platelets, erythrocytes, and reticulocytes. Furthermore, the training samples used to train the deep learning neural network model in this specific embodiment differ from those in Specific Embodiment 1-1; the training samples in this specific embodiment are blood cell scatter distribution images including platelets, erythrocytes, and reticulocytes.

[0097] like Figure 5 As shown, a scatter image of blood cell distribution, composed of forward-scattered light signals and lateral fluorescence signals, is input into a deep learning neural network model. After downsampling and upsampling, the deep learning neural network model outputs the cell classification results of the scatter image, indicating the particle clusters corresponding to the three cell types: platelets, erythrocytes, and reticulocytes. The deep learning neural network model used in this embodiment can have the same structure as that used in specific embodiment 1-1. For a description of this model, please refer to the above description; it will not be repeated here.

[0098] In another implementation, the detection device 130 of the blood cell analysis system is an impedance detection device. The data processing device 140 obtains blood cell impedance data based on the impedance pulse signal collected by the impedance detection device, and inputs the blood cell impedance data into a deep learning neural network model to obtain the classification label corresponding to the particles in the blood cell impedance data. The classification label is used to represent the cell type of the particles.

[0099] It should be noted that if the blood cell impedance data is in the form of blood cell histograms, the deep learning neural network model adds a corresponding classification label to each blood cell histogram. This classification label indicates the cell type of the blood cell particles corresponding to that blood cell histogram. If the blood cell impedance data is in the form of blood cell histogram images, the deep learning neural network model adds classification labels to the blood cell histogram images. These classification labels indicate the cell type of the blood cell particles. For example, in a one-dimensional blood cell histogram image, demarcation lines are used to indicate the classification of different cell types.

[0100] Specific Examples 1-4

[0101] A blood analysis system includes a sampling device, a sample preparation device, an impedance detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce the test sample solution. Each blood cell in the test sample solution passes through the impedance detection device to obtain the impedance signal of the blood cells, which can be specifically represented as impedance pulse signals. The data processing device converts the impedance pulse signals of the blood cells into a blood cell histogram for platelets and red blood cells, where the horizontal axis represents the blood cell volume and the vertical axis represents the number of blood cells. The blood cell histogram can be called a one-dimensional histogram, a volume statistics histogram, a one-dimensional histogram sequence, or a one-dimensional sequence.

[0102] To improve the accuracy of deep learning neural network models in classifying cells within one-dimensional histograms, data processing devices can perform preprocessing operations such as filtering, denoising, and enhancement on the one-dimensional histograms before inputting the preprocessed histograms into the trained deep learning neural network model. If the training samples for the deep learning neural network model are one-dimensional histograms of blood cells, including platelets and red blood cells, then the trained deep learning neural network model can determine the corresponding blood cell volume ranges for platelets and red blood cells within that one-dimensional histogram. For example... Figure 6A As shown, the one-dimensional histograms of platelets and red blood cells are input into a deep learning neural network model. After feature extraction and processing, the deep learning neural network model outputs the cell classification results of the one-dimensional histogram, that is, cells with a volume range of 0 to 50 are platelets (PLT), and cells with a volume range of 50 to 250 are red blood cells (RBC). Figure 6A The dashed box in the middle represents a deep learning neural network model.

[0103] For example, deep learning neural network models can specifically include LSTM, GRU, etc. See Figure 6B Taking an LSTM network model as an example, the structure and classification steps of this deep learning neural network model are explained. The model includes an LSTM layer (comprising multiple layers, such as 100 hidden units), a fully connected layer, a softmax layer, and a classification layer. After a one-dimensional histogram is input into the LSTM network model, the classification result is obtained sequentially through each layer. This deep learning neural network model can be applied to multiple embodiments of this application to identify various cell types included in the one-dimensional histogram. In this embodiment, the model identifies platelets and red blood cells.

[0104] Specific embodiments 1-5

[0105] The blood analysis system includes a sampling device, a sample preparation device, an impedance detection device, and a data processing device. In the reaction chamber of the sample preparation device, the processing reagents react with the blood sample obtained from the sampling device to produce the test sample solution.

[0106] Similar to embodiments 1-4, the blood cells in the sample solution are passed through an impedance detection device one by one to obtain impedance pulse signals. The data processing device converts the pulse signals of the blood cells into one-dimensional histograms of lymphocytes, intermediate cells, and neutrophils, and inputs the preprocessed one-dimensional histograms into a deep learning neural network model. Unlike embodiments 1-1, the deep learning neural network model outputs the corresponding blood cell volume ranges for lymphocytes, intermediate cells, and neutrophils in the one-dimensional histogram. Furthermore, the training samples used to train the deep learning neural network model in this embodiment differ from those in embodiments 1-4; the training samples in this embodiment are one-dimensional histograms including lymphocytes, intermediate cells, and neutrophils. It should be noted that intermediate cells refer to the collective term for three systems: monocytes, eosinophils, and basophils.

[0107] like Figure 7 As shown, the one-dimensional histograms of lymphocytes (LYM), intermediate cells (MID), and neutrophils (GRAN) are input into a deep learning neural network model. After feature extraction and processing, the deep learning neural network model outputs the cell classification results of the one-dimensional histogram, that is, cells with a volume range of 0 to 45 are lymphocytes (LYM), cells with a volume range of 45 to 65 are intermediate cells (MID), and cells with a volume range of 65 to 175 are neutrophils (GRAN).

[0108] The second type of blood cell analysis system can output statistical results of cell types.

[0109] In one implementation, the detection device 130 of the blood cell analysis system is an optical detection device. The data processing device 140 obtains blood cell scatter distribution data based on the optical pulse signals collected by the optical detection device, and inputs the blood cell scatter distribution data into a deep learning neural network model to obtain statistical results of the cell types included in the blood cell scatter distribution data. For example, the statistical results of cell types can be the proportion of blood cells, such as the ratio between different cell types, or the proportion of the number of cells of each cell type to the total number of blood cells in the sample solution to be tested.

[0110] Specific Examples 1-6

[0111] The blood analysis system includes a sampling device, a sample preparation device, a hemoglobin detection device, and a data processing device. In the reaction chamber of the sample preparation device, the processing reagents react with the blood sample obtained from the sampling device to produce the test sample solution.

[0112] The sample solutions are passed one by one through a hemoglobin detection device, which specifically detects changes in the transmittance of the sample solutions. A data processing unit generates a hemoglobin voltage curve based on the changes in transmittance and inputs this curve into a deep learning neural network model. The deep learning neural network model then determines the hemoglobin concentration in the sample solution based on the hemoglobin voltage curve. One structure of the deep learning neural network model is shown below. Figure 8 As shown, it may include LSTM layers, FC-256 layers, FC-64 layers, FC-1 layers, and regression layers. It should be noted that the deep learning neural network model can also have other structures, and this application does not impose specific limitations on these structures.

[0113] Specific embodiments 1-7

[0114] The blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. The processing reagents may include hemolysins and fluorescent agents. In the reaction chamber of the sample preparation device, these reagents react with the blood sample obtained from the sampling device to produce a test sample solution. Blood cells in the test sample solution pass through the optical detection device, generating lateral scattered light and fluorescence signals. The data processing device converts the scattered light and fluorescence signals from the blood cells into a scatter distribution image of the blood cells. This scatter distribution image is then input into a deep learning neural network model, which outputs statistical results, such as the counts of neutrophils, eosinophils, basophils, lymphocytes, and monocytes, or the count of white blood cells, or the count of immature granulocytes.

[0115] like Figure 9 As shown, one structure of the deep learning neural network model includes VGG-Net, FC-4096, FC-1024, FC-6, and a regression layer. After inputting a scatter plot image of blood cells into the deep learning neural network model, the model can identify particle clusters corresponding to six cell types in the image: neutrophils, eosinophils, basophils, lymphocytes, monocytes, and immature granulocytes, as well as the proportion of each cell type in the sample solution. It should be noted that... Figure 9 The model may not have a solid coil (classification label), meaning that the deep learning neural network model may not output a classification label, but can directly output the count value of each cell type.

[0116] It should be noted that, in this embodiment, the data processing device may input the deep learning neural network model not with an intuitive blood cell scatter plot image, but with blood cell scatter plot data that generates the blood cell scatter plot image. The deep learning neural network model then obtains the statistical results of cell types based on the blood cell scatter plot data.

[0117] As can be seen from the above specific embodiments of the blood analysis system, the blood analysis system can classify blood cells, such as outputting the classification labels corresponding to particles, or directly outputting statistical results of blood cell types.

[0118] Application Scenario 2: Correcting cell distribution data.

[0119] Blood analysis systems can detect blood cell distribution data, but this data may be inaccurate due to bias or error. To improve the accuracy of the blood cell distribution data, correction operations can be performed. Therefore, this application provides a blood analysis system, which includes at least a sampling device, a sample preparation device, a detection device, and a data processing device. It should be noted that descriptions of the first three devices can be found in the relevant content of Application Scenario 1, and will not be repeated here. The following mainly describes the structure that differs from Application Scenario 1: the data processing device.

[0120] A data processing device, connected to a detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program. When the computer program is invoked by the processor, the following steps are executed: generating blood cell distribution data based on the pulse signals of the cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain corrected blood cell distribution image data output by the deep learning neural network model. It should be noted that if the blood cell distribution data is an image, the correction can make the boundaries between particle clusters in the image clearer; if the blood cell distribution data is a data array, the correction method can be to modify the data values ​​of the blood cell distribution data to make the data array obtained from the pulse signals more accurate.

[0121] Specifically, the sampling device, sample preparation device, and detection device can obtain the pulse signals of cells in the sample solution to be tested. The data processing device first generates blood cell distribution data from the cell pulse signals. In order to distinguish it from the blood cell distribution data after correction processing, the blood cell distribution data can be called the initial blood cell distribution data.

[0122] The data processing device can store a pre-trained deep learning neural network model, which is trained using blood cell distribution data training samples. Each blood cell distribution data training sample includes the blood cell distribution data and the corrected blood cell distribution data. The deep learning neural network model trained by these training samples has the following correction capability: after the initial blood cell distribution data is input into the deep learning neural network model, the model corrects the initial blood cell distribution data. For example, the correction can remove interfering information such as cell debris, small cells, etc. Interfering information can affect the accuracy of the blood cell distribution results in the distribution map, such as unclear distribution boundaries, an excessive number of particle clusters, and an increased particle volume range in the one-dimensional histogram. After obtaining the initial blood cell distribution data, the data processing device inputs the initial blood cell distribution data into the aforementioned trained deep learning neural network model, which corrects the initial blood cell distribution data to obtain more accurate blood cell distribution data. The corrected blood cell distribution data can be referred to as the target blood cell distribution data. The distribution information in the target blood cell distribution data is more consistent with the actual situation of blood cells and has higher accuracy. The subsequent analysis and processing results based on the target blood cell distribution data are also more accurate.

[0123] The following provides specific embodiments of several blood analysis systems capable of performing the above functions.

[0124] Specific Implementation Example 2-1

[0125] The blood analysis system includes a sampling device, a sample preparation device, an impedance detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. Blood cells in the test sample solution pass one by one through the impedance detection device, which specifically detects the change in electrical impedance of the cells as they pass through the detection orifice, generating a pulse signal. The data processing device generates blood cell histogram data (or a one-dimensional histogram data array) based on the pulse signals from the cells and inputs this data into a deep learning neural network model to obtain the interference-reduced blood cell histogram data output by the deep learning neural network model.

[0126] For example, if the one-dimensional histogram data array is a platelet one-dimensional histogram data array, and the one-dimensional histogram data array may contain interference information, it can be called the initial one-dimensional histogram data array of platelets or the initial platelet one-dimensional histogram data array.

[0127] Blood analysis systems can provide classification results for platelets and red blood cells. For example... Figure 10AAs shown in the left-hand image, if the sample solution is relatively normal, there will be a significant difference in the volume of platelets and red blood cells. This difference can be distinguished by volume information such as size, thus not affecting the accuracy of subsequent data analysis of the two cell types. However, if the sample solution has abnormalities, such as… Figure 10A As shown in the right-hand figure, the sample solution contains a large number of small-volume red blood cells or red blood cell fragments. The volume classification method will misidentify these red blood cells as platelets, thus affecting the platelet data analysis results, such as causing a falsely high platelet count.

[0128] To eliminate interference information in the initial platelet one-dimensional histogram data array, a deep learning neural network model can be pre-trained. The trained deep learning neural network model has the following predictive capabilities: it can use deep learning methods to mine channel information from impedance detection channels and predict the platelet one-dimensional histogram data array after removing interference information through regression. One structure of the deep learning neural network model is shown below. Figure 10B As shown, the model includes Alex-Net layers, FC-4096, FC-4096, FC-256, and regression layers. The data processing device inputs the initial platelet one-dimensional histogram data array into the trained deep learning neural network model. This model can then remove interference from red blood cell pulse signals with volumes smaller than a volume threshold from the initial platelet one-dimensional histogram data array, thus obtaining a one-dimensional histogram data array containing only platelets, improving the accuracy of platelet statistical results. For ease of description, the interference-free initial platelet one-dimensional histogram data array can be called the target platelet one-dimensional histogram data array. In practical applications, various analyses and processing of platelets can be performed based on the target platelet one-dimensional histogram data array according to processing requirements, including but not limited to platelet counting.

[0129] See Figure 10C An example of interference cancellation is provided. Line 1 represents the initial platelet one-dimensional histogram obtained from the initial platelet one-dimensional histogram data array, and line 2 represents the target platelet one-dimensional histogram, which is obtained by transforming the target platelet one-dimensional histogram data array. The target platelet one-dimensional histogram data array is obtained by a deep learning neural network model after removing interference information from the initial platelet one-dimensional histogram data array. The horizontal axis of the histogram represents the blood cell volume, and the dashed line indicates a boundary with a specific volume value. The left and right sides of lines 1 and 2 represent the statistical results of platelets and red blood cells, respectively. By comparing the statistical results of platelets on the left side of the dashed lines, it can be seen that due to the influence of small red blood cells and red blood cell fragments, the statistical results of the initial platelet one-dimensional histogram are higher than those of the target platelet one-dimensional histogram, as shown by the shaded diagonal lines. The area difference of this part of the two lines represents the interference information caused by red blood cell fragments or small red blood cells.

[0130] Specific Implementation Examples 2-2

[0131] A blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. Blood cells in the test sample solution pass through the optical detection device, generating at least two types of optical pulse signals. The data processing device generates a blood cell scatter plot data from these signals and inputs it into a deep learning neural network model to obtain corrected blood cell scatter plot data. For ease of description, the uncorrected blood cell scatter plot data is referred to as the initial blood cell scatter plot data.

[0132] Furthermore, the data processing device inputs the initial blood cell scatter plot data into a pre-trained deep learning neural network model. The deep learning neural network model then performs correction operations on the plot data to make it more accurate. The processed plot data can be referred to as the target blood cell scatter plot data. See [link / reference] Figure 11 As shown in the example of correction, the initial blood cell scatter plot data can be converted into an intuitive initial blood cell scatter plot distribution image. This image is generated based on side-scattered light signals (SS) and forward-scattered light signals (FS). The deep learning neural network model includes an FCN layer and a regression layer. In the target blood cell scatter plot distribution image obtained after correction processing by the deep learning neural network model, the boundaries between cell clusters are clearer. The target blood cell scatter plot distribution image is converted from the corrected blood cell scatter plot data. In practical applications, various analytical processing operations can be performed based on the corrected blood cell scatter plot data, such as cell cluster classification and cell counting, depending on the processing requirements.

[0133] In the above application scenarios, before inputting the blood cell scatter plot data into the deep learning neural network model, the data processing device can further perform preprocessing operations on the blood cell scatter plot data. The preprocessing operations include at least one of filtering, denoising, and enhancement to improve the accuracy of the processing results of the deep learning neural network model on the blood cell scatter plot data.

[0134] Application Scenario 3: Identifying the sample type of a blood sample.

[0135] A blood analysis system can obtain blood cell detection data from a blood sample, such as blood cell distribution data and blood cell result parameters. Based on the blood cell detection data, the characteristics of the blood sample can be further determined, thereby determining the sample type. Therefore, this application provides a blood analysis system, which includes at least a sampling device, a sample preparation device, a detection device, and a data processing device. The sampling device, sample preparation device, and detection device can obtain pulse signals from the cells in the sample solution to be tested. The data processing device first generates blood cell detection data from the cell pulse signals. It should be noted that the descriptions of the first three devices can be found in the relevant content of Application Scenario 1, and will not be repeated here. The following mainly describes the structure that differs from Application Scenario 1: the data processing device.

[0136] A data processing device, connected to a detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program. When the computer program is invoked by the processor, the following steps are executed: generating blood cell detection data based on cell pulse signals. It should be noted that, compared to the data processing device in application scenario one, the blood cell detection data obtained by the data processing device in this embodiment is richer, including not only blood cell distribution data but also blood cell result parameters such as statistical results and classification results. The data processing device further inputs the blood cell detection data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample. The sample type can include any one or more of the following: whether the blood sample is normal, whether the blood sample needs re-examination, the substance type of the blood sample, etc. As long as the deep learning neural network model is trained with training samples of a certain sample type, the deep learning neural network model can detect and analyze the characteristics of the blood sample in any aspect and determine the sample type of the blood sample. The sample type can be output by the deep learning neural network model in the form of a type label.

[0137] The data processing device can store a pre-trained deep learning neural network model, which is trained using blood cell test data training samples. The deep learning neural network model trained with these training samples has the following detection capability: it can detect the sample type of the blood sample.

[0138] Deep learning neural network models can include, but are not limited to, CNN models. For example... Figure 12The diagram illustrates an example structure of a deep learning neural network model, including an input layer, a 5×5 conv-32 layer, a ReLU_1 layer, a Max pooling layer, a 5×5 conv-32 layer, a ReLU_2 layer, a Max pooling layer, a 5×5 conv-64 layer, a ReLU_1 layer, a Max pooling layer, an FC-64 layer, an FC-2 layer, a SoftMax layer, and an output layer. Deep learning neural network models can extract graph features, primarily through convolutional operations performed by multiple convolutional layers. Due to different parameter distributions, convolutional layers can respond differently to local features. For example, the first convolutional layer might have a high response to low-level features of the input data, such as image contours. As features from lower convolutional layers propagate forward, higher convolutional layers further extract and merge the features extracted by lower layers into higher-level semantic information. ReLU is the activation function used by the model, mainly to improve the model's non-linear expressive power. Max pooling layers are used to downsample and reduce the dimensionality of the information extracted by the model and remove redundant information, thereby simplifying model complexity, reducing computation, and increasing the translation invariance of features. Fully connected layers (FC layers) concatenate the extracted local feature information to serve as voting values ​​for classification decisions, and the output layer ultimately determines the sample type.

[0139] After obtaining the blood cell test data, the data processing device inputs the blood cell test data into the deep learning neural network model that has been trained, and the model obtains the sample type based on the blood cell test data.

[0140] The following provides specific embodiments of several blood analysis systems capable of performing the above functions.

[0141] Specific Implementation Example 3-1

[0142] The blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. The blood cells in the test sample solution pass one by one through the optical detection device, obtaining at least two optical pulse signals from the blood cells, such as forward-scattered light pulse signals, side-scattered light pulse signals, and fluorescence pulse signals. It should be noted that at least one of these at least two optical pulse signals is a fluorescence pulse signal.

[0143] The data processing device converts optical pulse signals into blood cell scatter distribution data, such as two-dimensional or three-dimensional scatter distribution data, including fluorescence intensity data. The data processing device further inputs the blood cell scatter distribution data into a pre-trained deep learning neural network model to obtain the blood sample type. The sample type indicates whether the blood sample is normal. Specifically, the sample type includes one of the following: the number of primitive cells in the blood sample exceeds a preset threshold, or the number of primitive cells in the blood sample does not exceed a preset threshold; the number of immature cells in the blood sample exceeds a preset threshold, or the number of immature cells in the blood sample does not exceed a preset threshold; the blood sample is infected with Plasmodium, or the blood sample is not infected with Plasmodium.

[0144] Based on the sample type predicted by the deep learning neural network model, it can be clearly determined whether a blood sample is normal or abnormal. For example, an abnormality could specifically be due to an excess of primitive cells, an excess of immature cells, or infection by malaria parasites. If an abnormality is detected, a corresponding alarm can be displayed, such as a primitive alarm, an immature cell alarm, or a malaria alarm.

[0145] One specific implementation method involves an optical detection device acquiring forward-scattered light, side-scattered light signals, and fluorescence signals from blood cells in the sample fluid. A data processing device then processes these signals into a three-dimensional scatter distribution image of blood cells. A two-dimensional scatter distribution image of blood cells from a specific viewpoint is selected from the three-dimensional image and input into a deep learning neural network model. If the number of primitive cells in the sample fluid exceeds a threshold, the deep learning neural network model outputs the corresponding type label; otherwise, it outputs the corresponding type label.

[0146] like Figure 13A As shown, the data processing device selects a two-dimensional blood cell scatter image from the SSC_SFL perspective from a three-dimensional blood cell scatter image composed of side-scattered light signals (SS), side-fluorescence scattered light signals (FL), and forward-scattered light signals (FS). Figure 13B As shown in the upper middle figure, if the number of original cells in the blood cell scatter distribution image is excessive, the CNN neural network will output a marker 1 to indicate that the original cells in the blood cell scatter distribution image are abnormal, and may also display an alarm. As shown in the lower middle figure of 13B, if the number of original cells in the blood cell scatter distribution image is not excessive, the CNN neural network will output a marker 0 to indicate that the original cells in the blood cell scatter distribution image are normal, and may also display that the detection result is normal.

[0147] Specific Implementation Example 3-2

[0148] The blood analysis system includes a sampling device, a sample preparation device, an optical detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. The blood cells in the test sample solution pass one by one through the optical detection device, obtaining at least two optical pulse signals for the blood cells, such as forward-scattered light pulse signals, side-scattered light pulse signals, and fluorescence pulse signals. It should be noted that the fluorescence pulse signal is not included among these at least two optical pulse signals.

[0149] In practical applications, some blood analysis systems may lack fluorescence detectors, preventing optical detection devices from acquiring fluorescence information. To enable such systems to output original alarms, the deep learning neural network model used by the data processing device can be pre-trained. Specifically, blood cell detection data, including fluorescence signals, is used to label training samples, such as indicating whether original cells exceed a threshold or not. These labeled training samples are then input into the deep learning neural network model for training, enabling the trained model to output whether original cells exceed the threshold.

[0150] Therefore, after the detection device of the aforementioned type of blood analysis system receives the pulse signal of the sample solution, the data processing device generates blood cell scatter distribution data based on the cell pulse signal. This scatter distribution data does not include fluorescence intensity data. For example, taking the BC-5390CRP hematology analyzer manufactured by Mindray as an example, the blood cell scatter distribution data generated by the data processing device includes data from both forward and side-scatter signals, but excludes fluorescence intensity data. The data processing device further inputs this blood cell scatter distribution data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample. The sample type is either: the amount of primitive cells in the blood sample exceeds a preset threshold, or the amount of primitive cells in the blood sample does not exceed a preset threshold. Figure 13C As shown in the upper middle figure, if the blood cell scatter distribution image contains an excessive number of primitive cells, the CNN neural network will output a marker 1 to indicate that the primitive cells in the blood cell scatter distribution image are abnormal, and may also display a primitive alarm; as shown in the lower middle figure of 13C, if the blood cell scatter distribution image does not contain an excessive number of primitive cells, the CNN neural network will output a marker 0 to indicate that the primitive cells in the blood cell scatter distribution image are normal, and may also display that the detection result is normal.

[0151] Alarm information from abnormal samples plays a crucial role in clinicians' diagnosis. Based on these alarm parameters, along with indicators such as the number, morphology, and distribution of cells in the patient, doctors can quickly and effectively identify various blood diseases. Therefore, the accuracy of the sample type information provided by the blood analyzer is critical to a doctor's ability to make accurate judgments. Because two-dimensional scatter images of cells lack rich texture information, traditional image algorithms struggle to extract features, leading to frequent misleading or false alarms from existing blood analyzers. This severely impacts doctors' accurate disease diagnosis and the optimal timing of patient treatment. The sample type results obtained through a deep learning neural network model in this application are more accurate, effectively reducing the alarm error rate and improving the accuracy of disease diagnosis.

[0152] Specific Implementation Examples 3-3

[0153] A blood analysis system includes a sampling device, a sample preparation device, a detection device, and a data processing device. In the reaction chamber of the sample preparation device, the processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. Each blood cell in the test sample solution is then passed through the detection device to obtain a pulse signal from the blood cells. It should be noted that the detection device can obtain multiple different types of pulse signals from the blood sample. For example, different processing reagents can be used to prepare multiple different test sample solutions from the same blood sample, and these multiple test sample solutions can obtain multiple different types of pulse signals when passed through the same detection device; alternatively, the same processing reagents can be used to prepare multiple different test sample solutions from the same blood sample, and these multiple test sample solutions can obtain multiple different types of pulse signals when passed through multiple different detection devices; or a combination of these methods.

[0154] The data processing device generates various types of blood cell detection data based on different types of pulse signals from the blood sample. This includes generating two-dimensional or three-dimensional blood cell scatter plot data, blood cell histogram data based on pulse signals, and blood cell parameter results based on scatter plot data and histogram data. These different types of blood cell detection data are then input into a pre-trained deep learning neural network model. The model determines the sample type of the blood sample: whether it requires retesting or does not require retesting (one-time testing). A blood sample that does not require retesting indicates that it is normal.

[0155] like Figure 14AAs shown, after the sample solution is passed through the DIFF channel, RET channel, and WNB channel, the detection device can further obtain scatter images of blood cell distribution and histograms of platelets and red blood cells from different perspectives of each channel based on the forward scatter intensity, side scatter intensity, fluorescence intensity, and impedance voltage curves. The data processing device can input all or part of the blood cell detection data into a deep learning neural network model for retesting and prediction, such as... Figure 14B As shown, an abnormal blood cell scatter distribution image and a normal blood cell scatter distribution image from the DIFF channel can be input into a CNN neural network respectively. The CNN neural network will output a label of 1 for the abnormal blood cell scatter distribution image and a label of 0 for the normal blood cell scatter distribution image; for example... Figure 14C As shown, an abnormal blood cell scatter distribution image and a normal blood cell scatter distribution image from the RET channel can be input into a CNN neural network, respectively. The CNN neural network can output labels 1 and 0 for the abnormal blood cell scatter distribution image and the normal blood cell scatter distribution image, respectively; for example... Figure 14D As shown, an abnormal blood cell scatter plot image and a normal blood cell scatter plot image from the WNB channel can be input into a CNN neural network. The CNN neural network can output labels 1 and 0 for the abnormal blood cell scatter plot image and the normal blood cell scatter plot image, respectively. Here, label 0 indicates that the blood cell scatter plot image is normal, and label 1 indicates that the blood cell scatter plot image needs to be re-examined.

[0156] Blood cell distribution data can be further processed to obtain the parameter results of the blood sample. These parameter results can be presented in tabular form. Table 2 provides a specific example of a table showing the parameter results of a blood sample.

[0157] Table 2

[0158]

[0159] The parameters of the blood sample are input into a deep learning neural network model, such as a backpropagation (BP) neural network. Based on the results of each detection parameter, the neural network can output the sample type of whether the blood sample needs to be retested. For example, if the output label is 1, it means that the blood sample needs to be retested, and if the output label is 0, it means that the blood sample does not need to be retested.

[0160] It should be noted that the blood cell distribution data and the result parameters of the blood sample can be input into the deep learning neural network model separately to detect whether a retest is needed, or they can be input into the model together to detect whether a retest is needed.

[0161] Specific embodiments 3-4

[0162] In addition to detecting blood samples, blood analysis systems can also detect other types of substances. When the type of substance to be tested is unknown, the blood analysis system is needed to determine the sample type.

[0163] A blood analysis system includes a sampling device, a sample preparation device, a detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce a test sample solution. Blood cells in the test sample solution are individually detected by an optical detection device to obtain optical pulse signals. The data processing device generates scatter plot data of blood cells based on the optical pulse signals and inputs this data into a pre-trained deep learning neural network model to determine the sample type of the blood sample; the sample type is either a blood sample, a quality control sample, or a calibrator. It should be noted that quality control samples and calibrators are names for two sample types in the medical field.

[0164] It should be noted that the samples collected by the sampling device may be blood or other substances. The deep learning neural network model is pre-trained using labeled blood cell scatter distribution data as training samples. The labels include blood sample type, quality control material type, and calibrator type. After training the deep learning neural network model using these training samples, the model gains the ability to distinguish between different sample types. Deep learning neural network models include, but are not limited to, CNN neural networks.

[0165] In one specific embodiment, blood cells in the sample fluid to be tested pass through an optical detection device one by one to obtain forward-scattered light, side-scattered light signals, and fluorescence signals for each blood cell in the sample fluid. The data processing device can obtain scatter distribution data of blood cells from various viewpoints based on the forward-scattered light, side-scattered light signals, and fluorescence signals. It can then select scatter distribution data from a specific viewpoint or combine scatter distribution data from multiple different viewpoints and input it into a deep learning neural network model. The deep learning neural network model then outputs the sample type of the sample fluid to be tested (i.e., the blood sample).

[0166] like Figure 15 As shown, the data processing device inputs three blood cell scatter distribution images from the SSC_FL perspective into the CNN neural network. If the CNN neural network outputs label 1, it indicates that the blood sample corresponding to the blood cell scatter distribution image is a blood sample. If the CNN neural network outputs label 2, it indicates that the blood sample corresponding to the blood cell scatter distribution image is a quality control material. If the CNN neural network outputs label 3, it indicates that the blood sample corresponding to the blood cell scatter distribution image is a calibrator.

[0167] In the above application scenarios, the blood analysis system embodiment allows the data processing device to determine the sample type of a blood sample based on blood cell detection data.

[0168] Application Scenario 4: Detecting whether there are any abnormalities in the sample measurement process.

[0169] Blood analysis systems may encounter anomalies during blood sample measurement, such as blocked or improper aspiration. These anomalies can cause the system to malfunction or result in inaccurate measurements. Therefore, it is necessary to provide alerts indicating whether the blood sample measurement process is abnormal. To address this, this application provides a blood analysis system comprising at least a sampling device, a sample preparation device, a detection device, and a data processing device. It should be noted that descriptions of the first three devices can be found in the relevant content of Application Scenario 1, and will not be repeated here. The following mainly describes the structure that differs from Application Scenario 1: the data processing device.

[0170] A data processing device, connected to a detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: generating blood cell distribution data based on cell pulse signals, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain a prompt indicating whether the blood sample measurement process is abnormal.

[0171] The blood cell distribution data can be curve data obtained from the pulse signals of blood cells. Abnormalities can include, but are not limited to, sampling abnormalities or well blockage. Well blockage may be caused by an abnormality in the blood sample collection device or by existing blockage in the blood itself; the specific cause can be determined through further analysis. Abnormality alerts can be displayed in various ways, such as by showing a message on a monitor or by an alarm device outputting an alarm message. The deep learning neural network model can be a CNN neural network model; the structure of this model can be found in the above description and will not be repeated here.

[0172] The blood analysis system will be described in detail below through several different specific embodiments.

[0173] Specific Implementation Example 4-1

[0174] The blood analysis system includes a sampling device, a sample preparation device, an impedance detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce the test sample solution. Blood cells in the test sample solution pass one by one through the impedance detection device, which detects the change in electrical impedance of the cells passing through the detection orifice to obtain an impedance voltage pulse signal.

[0175] The data processing device preprocesses the cell impedance voltage pulse signal through noise reduction and filtering to generate impedance voltage curve data (or simply impedance curve data). This data is then input into a deep learning neural network model, such as a CNN, which can output a prompt indicating whether the blood sample measurement process is clogged. The presence of clogged pores indicates an abnormality in the sample fluid (i.e., the blood sample).

[0176] like Figure 16 As shown in the image above, after inputting the impedance-voltage curve (a visual representation of impedance-voltage curve data) corresponding to a normal blood sample into a CNN neural network, the CNN neural network will output a marker of 0 to indicate that the blood sample is normal. Figure 16 As shown in the figure below, after the impedance voltage curve generated by a blood sample with blocked aperture or abnormal sampling is input into a CNN neural network, the CNN neural network will output a marker 1 to indicate that the blood sample has blocked aperture.

[0177] Specific Implementation Example 4-2

[0178] The blood analysis system includes a sampling device, a sample preparation device, an impedance detection device, and a data processing device. In the reaction chamber of the sample preparation device, processing reagents react with the blood sample obtained from the sampling device to produce the test sample solution. The blood cells in the test sample solution pass one by one through the impedance detection device, which detects the change in electrical impedance of the cells passing through the detection orifice to obtain the aspiration pressure pulse signal.

[0179] The data processing device preprocesses the cell impedance voltage pulse signal by denoising and filtering to generate aspiration pressure curve data. The aspiration pressure curve data is then input into a pre-trained deep learning neural network model, such as a CNN neural network, to obtain a prompt indicating whether there is an aspiration abnormality during the blood sample measurement process.

[0180] like Figure 17 As shown in the image above, after inputting the aspiration pressure curve (a visual representation of aspiration pressure curve data) corresponding to a normal blood sample into a CNN neural network, the CNN neural network will output a marker of 0 to indicate that the blood sample is normal. Figure 17 As shown in the figure below, after the aspiration pressure curve generated from a blood sample with aspiration abnormality is input into the CNN neural network, the CNN neural network will output a label 1 to indicate that the blood sample has aspiration abnormality.

[0181] The blood analysis systems described above allow the data processing device to input the blood cell distribution data corresponding to the blood sample into a deep learning neural network model. The model then analyzes the blood cell distribution data to detect any abnormalities in the blood sample measurement process and provides feedback on the test results.

[0182] The features described above in the disclosed embodiments can be substituted or combined with each other to enable those skilled in the art to implement or use this application. This document has been described with reference to various exemplary embodiments. However, those skilled in the art will recognize that changes and modifications can be made to the exemplary embodiments without departing from the scope of this document. For example, various operational steps and components for performing operational steps can be implemented in different ways depending on the specific application or considering any number of cost functions associated with the operation of the system (e.g., one or more steps can be deleted, modified, or combined with other steps).

[0183] The terms "first," "second," etc., used in this specification, claims, and the accompanying drawings are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, or apparatuses.

[0184] Furthermore, as those skilled in the art will understand, the principles herein can be reflected in a computer program product on a computer-readable storage medium pre-loaded with computer-readable program code. Any tangible, non-transitory computer-readable storage medium may be used, including magnetic storage devices (hard disks, floppy disks, etc.), optical storage devices (CD-ROMs, DVDs, Blu-ray discs, etc.), flash memory, and / or the like. These computer program instructions may be loaded onto a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to form a machine, such that instructions, which execute on the computer or other programmable data processing apparatus, can generate means for performing a specified function. These computer program instructions may also be stored in a computer-readable storage medium that can instruct the computer or other programmable data processing apparatus to operate in a particular manner, such that instructions stored in the computer-readable storage medium can form an article of manufacture, including means for implementing the specified function. The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to perform a series of operational steps on the computer or other programmable apparatus to produce a computer-implemented process, such that instructions, which execute on the computer or other programmable apparatus, can provide steps for implementing the specified function.

[0185] The foregoing specific descriptions have been described with reference to various embodiments. However, those skilled in the art will recognize that various modifications and changes can be made without departing from the scope of this disclosure. Therefore, considerations for this disclosure are to be illustrative rather than restrictive, and all such modifications are to be included within its scope. Similarly, advantages, other advantages, and solutions to problems with respect to various embodiments have been described above. However, benefits, advantages, solutions to problems, and any elements that produce these, or make them more explicit, should not be construed as critical, essential, or necessary. The term “comprising” and any other variations thereof as used herein are non-exclusive inclusion, meaning that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed or not part of the process, method, system, article, or apparatus. Furthermore, the term “coupled” and any other variations thereof as used herein refer to physical connections, electrical connections, magnetic connections, optical connections, communication connections, functional connections, and / or any other connections.

[0186] The above embodiments are merely illustrative of several implementation methods, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this invention, and these all fall within the protection scope of this invention. Therefore, the protection scope of this invention should be determined by the appended claims.

Claims

1. A blood analysis system, characterized in that, include: A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device; A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested; A detection device is used to detect the sample solution to obtain pulse signals of multiple cells in the sample solution; A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: processing the pulse signals of the plurality of cells according to the pulse signals of the plurality of cells to obtain blood cell distribution data including population distribution information of the plurality of cells; inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain the blood cell classification result output by the deep learning neural network model; wherein the deep learning neural network model is trained by training samples, the training samples including blood cell distribution data samples and blood cell classification result labels corresponding to the blood cell distribution data samples.

2. The blood analysis system as described in claim 1, characterized in that, The detection device is an optical detection device; The cells in the sample solution to be tested pass through the optical detection device one by one, and the optical detection device collects at least two types of optical pulse signals from the cells. When the data processing device performs the step of obtaining blood cell distribution data based on the pulse signals of the cells, it is specifically used to obtain blood cell scatter distribution data based on at least two types of optical pulse signals of the cells.

3. The blood analysis system as described in claim 2, characterized in that, The data processing device is specifically used to input the blood cell scatter distribution data into the deep learning neural network model to obtain the classification label corresponding to the cell in the blood cell scatter distribution data, and the classification label is used to represent the cell type.

4. The blood analysis system as described in claim 3, characterized in that, The category label includes at least one of the following: Classification labels for neutrophils, eosinophils, basophils, lymphocytes, and monocytes; or, Classification labels for basophils, white blood cells, and optional nucleated red blood cells; or, Classification labels for platelets, red blood cells, and reticulocytes.

5. The blood analysis system as described in claim 2, characterized in that, The data processing device is specifically used to input the blood cell scatter distribution data into the deep learning neural network model to obtain statistical results of the cell types included in the blood cell scatter distribution data.

6. The blood analysis system as described in claim 5, characterized in that, The statistical results for the cell types include at least one of the following: The count values ​​of neutrophils, eosinophils, basophils, lymphocytes, and monocytes; or, White blood cell count; or, The count of immature granulocytes.

7. The blood analysis system as described in claim 2, characterized in that, When the data processing device performs the step of obtaining blood cell distribution data based on the pulse signals of the cells, it is specifically used to obtain blood cell scatter plot data based on at least two types of optical pulse signals of the cells.

8. The blood analysis system as described in claim 2, characterized in that, When the data processing device performs the step of obtaining blood cell distribution data based on the pulse signals of the cells, it is specifically used to obtain blood cell scatter plot data based on at least two types of optical pulse signals of the cells, and further process the blood cell scatter plot data to obtain a blood cell scatter distribution image.

9. The blood analysis system as described in claim 2, characterized in that, The optical detection device collects optical pulse signals in three dimensions: forward scattered light, side scattered light, and fluorescence intensity from the cells.

10. The blood analysis system as described in claim 1, characterized in that, The detection device is an impedance detection device; The impedance detection device is specifically used to detect the change in electrical impedance of the cell through the detection hole in the impedance detection device to obtain an impedance pulse signal; When the data processing device performs the step of obtaining blood cell distribution data based on the pulse signal of the cell, it is specifically used to obtain blood cell impedance data based on the impedance pulse signal of the cell.

11. The blood analysis system as described in claim 10, characterized in that, The data processing device is specifically used to input the blood cell impedance data into the deep learning neural network model to obtain the classification labels of the cells in the blood cell impedance data, wherein the classification labels are used to represent the cell type; wherein the training samples used to train the deep learning neural network model specifically include blood cell impedance data samples and the classification labels of the blood cell impedance data samples.

12. The blood analysis system as described in claim 10, characterized in that, The data processing device is specifically used to input the blood cell impedance data into the deep learning neural network model to obtain statistical results of the cell types contained in the blood cell impedance data; wherein the training samples used to train the deep learning neural network model specifically include blood cell impedance data samples and statistical results of the cell types contained in the blood cell impedance data samples.

13. The blood analysis system as described in claim 11 or 12, characterized in that, The cell type includes at least one of the following: Red blood cells and platelets; or, Lymphocytes, intermediate cells, and neutrophils.

14. The blood analysis system as described in claim 1, characterized in that, The detection device is a hemoglobin detection device; The hemoglobin detection device is specifically used to detect changes in the transmittance of the sample solution to be tested. The data processing device is specifically used to obtain a hemoglobin voltage curve based on the change in transmittance, and input the hemoglobin voltage curve into a pre-trained deep learning neural network model to obtain the hemoglobin concentration value in the sample solution to be tested.

15. The blood analysis system as described in claim 1, characterized in that, The classification result is obtained by the deep learning neural network model through analysis and processing of the input blood cell distribution data, and is output by the deep learning neural network model.

16. A blood analysis system, characterized in that, include: A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device; A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested; A detection device is used to detect the sample solution to obtain pulse signals of multiple cells in the sample solution; A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: processing the pulse signals of the plurality of cells according to the pulse signals of the plurality of cells to obtain blood cell distribution data including population distribution information of the plurality of cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain corrected blood cell distribution data output by the deep learning neural network model; wherein the deep learning neural network model is trained by training samples, the training samples including blood cell distribution data samples and the corrected blood cell distribution data corresponding to the blood cell distribution data samples.

17. The blood analysis system as described in claim 16, characterized in that, The detection device is an impedance detection device; The impedance detection device is specifically used to detect the change in electrical impedance of the cell through the detection hole in the impedance detection device to obtain a pulse signal; The data processing device is specifically used to obtain blood cell histogram data based on the pulse signal of the cells, input the blood cell histogram data into the deep learning neural network model, and obtain the interference-reduced blood cell histogram data output by the deep learning neural network model; wherein the training samples used to train the deep learning neural network model specifically include blood cell histogram data samples and the interference-reduced blood cell histogram data corresponding to the blood cell histogram data samples.

18. The blood analysis system as described in claim 16, characterized in that, The detection device is an optical detection device; The cells in the sample solution to be tested pass through the optical detection device one by one, and the optical detection device collects at least two types of optical pulse signals from the cells. The data processing device is specifically used to obtain blood cell scatter plot data based on at least two types of optical pulse signals from the cells, and input the blood cell scatter plot data into the deep learning neural network model to obtain corrected blood cell scatter plot data; wherein the training samples used to train the deep learning neural network model specifically include blood cell scatter plot data samples and the corrected blood cell scatter plot data corresponding to the blood cell scatter plot data samples.

19. A blood analysis system, characterized in that, include: A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device; A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested; A detection device is used to detect the sample solution to obtain pulse signals of multiple cells in the sample solution; A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: processing the pulse signals of the plurality of cells according to the pulse signals of the plurality of cells to generate blood cell detection data, and inputting the blood cell detection data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample.

20. The blood analysis system as described in claim 19, characterized in that, The detection device is an optical detection device; The data processing device is specifically used to generate blood cell scatter distribution data based on the pulse signals of the cells. The blood cell scatter distribution data includes fluorescence intensity data. The blood cell scatter distribution data is input into a pre-trained deep learning neural network model to obtain the sample type of the blood sample. The sample type includes one of the following: The amount of primitive cells in the blood sample exceeds a preset threshold, or the amount of primitive cells in the blood sample does not exceed a preset threshold; The number of immature cells in the blood sample exceeds a preset threshold, or the number of immature cells in the blood sample does not exceed a preset threshold; The blood sample is infected with Plasmodium, or the blood sample is not infected with Plasmodium.

21. The blood analysis system as described in claim 19, characterized in that, The detection device obtains various types of pulse signals from cells in multiple different test sample solutions of the blood sample; The data processing device is specifically used to generate various types of blood cell detection data based on various types of pulse signals from the blood sample, and input the various types of blood cell detection data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample; the sample type is either the blood sample needs to be retested or the blood sample does not need to be retested.

22. The blood analysis system as described in claim 19, characterized in that, The detection device is an optical detection device; The data processing device is specifically used to generate blood cell scatter distribution data based on the pulse signals of the cells, and input the blood cell scatter distribution data into a pre-trained deep learning neural network model to obtain the sample type of the blood sample; the sample type is blood sample, quality control material, or calibrator.

23. The blood analysis system as described in claim 19, characterized in that, The detection device is an optical detection device; The data processing device is specifically used to generate blood cell scatter distribution data based on the pulse signal of the cells. The blood cell scatter distribution data includes scattered light signal data but does not include fluorescence intensity data. The blood cell scatter distribution data is input into a pre-trained deep learning neural network model to obtain the sample type of the blood sample. The sample type is: the amount of primitive cells in the blood sample exceeds a preset threshold, or the amount of primitive cells in the blood sample does not exceed a preset threshold.

24. The blood analysis system as described in claim 19, characterized in that, The blood cell detection data includes: blood cell distribution data and / or result parameters of the blood sample.

25. A blood analysis system, characterized in that, include: A sampling device for collecting blood samples and transporting the blood samples to a sample preparation device; A sample preparation apparatus includes a reaction chamber and a reagent supply unit, wherein the reagent supply unit is used to provide processing reagents, and the reaction chamber is used to provide a reaction site for the blood sample and the processing reagents to react and prepare a sample solution to be tested; A detection device is used to detect the sample solution to obtain pulse signals of multiple cells in the sample solution; A data processing device, connected to the detection device, includes a processor and a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is invoked by the processor, the following steps are performed: processing the pulse signals of the plurality of cells according to the pulse signals of the plurality of cells to generate blood cell distribution data including population distribution information of the plurality of cells, and inputting the blood cell distribution data into a pre-trained deep learning neural network model to obtain a prompt indicating whether the measurement process of the blood sample is abnormal.

26. The blood analysis system as described in claim 25, characterized in that, The detection device is an impedance detection device; The impedance detection device is specifically used to detect the change in electrical impedance of the cell through the detection hole in the impedance detection device to obtain an impedance voltage pulse signal; The data processing device is specifically used to obtain impedance curve data based on the impedance voltage pulse signal of the cell, and input the impedance curve data into a pre-trained deep learning neural network model to obtain a prompt on whether the blood sample measurement process is blocked.

27. The blood analysis system as described in claim 25, characterized in that, The detection device is an impedance detection device; The impedance detection device is specifically used to detect the change in electrical impedance of cells through the detection orifice in the impedance detection device to obtain the sample aspiration pressure pulse signal; The data processing device is specifically used to obtain aspiration pressure curve data based on the aspiration pressure pulse signal of the cell, and input the aspiration pressure curve data into a pre-trained deep learning neural network model to obtain a prompt on whether there is an aspiration abnormality in the measurement process of the blood sample.

Images