Blood analysis system, method, device and computer storage medium

Abstract

Disclosed are a blood analysis system, method, device, and computer storage medium. The blood analysis system includes: a sample collection device for collecting a blood sample and transferring the blood sample to a sample preparation device; a sample preparation device for processing the blood sample using a reagent to prepare a sample solution to be tested; an impedance detection device for detecting the sample solution to be tested to obtain an impedance pulse signal of cells in the sample solution to be tested; and a data processing device for performing the following steps: obtaining a blood cell histogram of the sample solution to be tested based on the impedance pulse signal, inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, wherein the blood cell parameters are at least one of the following: a reticulocyte parameter, a reticulocyte platelet parameter, and a red blood cell fragment parameter; and the deep learning neural network model is obtained by training blood training samples labeled with blood cell parameters using a machine learning algorithm.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of blood testing, and more specifically, to a blood analysis system, method, device, and computer storage medium. Background Art

[0002] In order to analyze blood cells and determine various parameters related to the blood cells, a commonly used method currently is to obtain various parameters based on optical signal detection.

[0003] On the one hand, optical signal detection has high requirements for the environment. If there are factors in the environment that affect light transmission, the results will be inaccurate. On the other hand, optical signal detection requires multiple optical devices and other hardware, which is relatively expensive. Summary of the Invention

[0004] Embodiments of the present invention provide a blood analysis system, a blood analysis method, a blood analysis device, and a computer storage medium.

[0005] In a first aspect, a blood analysis system is provided, comprising:

[0006] A sample collection device, used for collecting a blood sample and transferring the blood sample to a sample preparation device;

[0007] A sample preparation device, used to process the blood sample using a reagent to prepare a sample liquid to be tested;

[0008] An impedance detection device, configured to detect the sample liquid to be tested to obtain an impedance pulse signal of cells in the sample liquid to be tested;

[0009] A data processing device is used to perform the following steps:

[0010] Obtaining a blood cell histogram of the sample liquid to be tested according to the impedance pulse signal, inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, wherein the blood cell parameters are at least one of the following: a reticulocyte parameter, a reticulocyte platelet parameter, and a red blood cell fragment parameter;

[0011] The deep learning neural network model is obtained by training blood training samples marked with blood cell parameters through a machine learning algorithm.

[0012] In a second aspect, a blood analysis method is provided, comprising:

[0013] Acquiring an impedance pulse signal of cells in the sample liquid to be tested obtained by an impedance detection device through detecting the sample liquid to be tested;

[0014] obtaining a blood cell histogram of the sample liquid to be tested according to the impedance pulse signal;

[0015] The blood cell histogram is input into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, where the blood cell parameters are at least one of the following:

[0016] Reticulocyte parameters, reticulocyte parameters and red blood cell fragment parameters;

[0017] The deep learning neural network model is obtained by training blood training samples marked with blood cell parameters through a machine learning algorithm.

[0018] In a third aspect, a blood analysis device is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the method of the second aspect when executing the computer program.

[0019] In a fourth aspect, a computer storage medium is provided, on which a computer program is stored, and when the computer program is executed by a computer or a processor, the steps of the method described in the second aspect are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 is a schematic block diagram of a blood analysis system according to an embodiment of the present invention;

[0022] Figure 2 is another schematic block diagram of a blood analysis system according to an embodiment of the present invention;

[0023] Figure 3 is a schematic diagram of an impedance detection device according to an embodiment of the present invention;

[0024] Figure 4 is a schematic block diagram of a data processing device according to an embodiment of the present invention;

[0025] Figure 5 is a schematic block diagram of a first deep learning network according to an embodiment of the present invention;

[0026] Figure 6 Schematic diagram of the correlation between RET% obtained in the embodiment of the present invention and RET% obtained by fluorescence staining;

[0027] Figure 7 Schematic diagram of the correlation between the IPF% obtained in the embodiment of the present invention and the IPF% obtained by fluorescent staining;

[0028] Figure 8 Schematic diagram of the correlation between the FRC% obtained in the embodiment of the present invention and the FRC% obtained by fluorescence staining;

[0029] Figure 9 is a schematic block diagram of a second deep learning network according to an embodiment of the present invention;

[0030] Figure 10 This is a schematic diagram of the volume difference between platelets and red blood cells in a normal sample;

[0031] Figure 11 Schematic diagram of the volume difference between platelets and red blood cells in the presence of small red blood cells and red blood cell fragments;

[0032] Figure 12 Schematic diagram comparing the PLT obtained in the embodiment of the present invention with the PLT obtained in the prior art;

[0033] Figure 13 is a schematic flow chart of a blood analysis method according to an embodiment of the present invention;

[0034] Figure 14 FIG. 4 is a schematic flow chart of a blood analysis method according to another embodiment of the present invention. DETAILED DESCRIPTION

[0035] The cells contained in human peripheral blood are usually mature blood cells. When the human body is diseased, immature cells will appear in the blood. The detection of immature cells can provide useful reference for the diagnosis and treatment of clinical diseases.

[0036] For example, reticulocyte (RET) and RBC fragment (FRC) are important clinical detection parameters that can reflect the red blood cell production function of the bone marrow. The immature platelet fraction (IPF) can be used as a diagnostic indicator for thrombocytopenia and one of the important indicators for observing therapeutic effects. Platelet count (PLT) plays an important role in helping clinicians determine whether a patient has a bleeding tendency. However, the conventional impedance method currently used to measure PLT will be interfered with by abnormal cells. For example, when there are red blood cell fragments or small-volume red blood cells in the sample, the impedance method will produce a falsely high PLT. When there are large-volume platelets in the sample, the impedance method will produce a falsely low PLT.

[0037] Current methods for detecting FRC, RET, and IPF in blood typically require adding dedicated detection channels to hematology analyzers, increasing instrument and reagent costs to achieve their counting function. For example, current methods can simultaneously count FRC, RET, and IPF in the RET channel.

[0038] The current detection method for PLT in the blood is similar. In order to solve the problem of impedance method being interfered by abnormal cells, an optical PLT-O channel is also added to detect PLT. Since optical PLT-O detection is not interfered with by abnormal cells, it can accurately report platelet count results.

[0039] However, the above-mentioned detection methods for various parameters all need to be implemented through optical signal detection methods, which are highly dependent on the environment and have high hardware costs.

[0040] The following will be combined Figures 1 to 14 Describe the specific implementation method in the embodiment of the present invention.

[0041] An embodiment of the present invention provides a blood analysis system for detecting blood cell parameters of a blood sample. The blood cell parameters may include, but are not limited to, at least one of networked red blood cell parameters, red blood cell fragment parameters, and platelet parameters. Networked red blood cell parameters include, but are not limited to, reticulocyte count (RET#) and reticulocyte count ratio (RET%). Red blood cell fragment parameters include, but are not limited to, red blood cell fragment count (FRC#) or red blood cell fragment ratio (FRC%). Platelet parameters include, but are not limited to, immature platelet fraction (IPF) and platelet count (PLT). The blood analysis system includes a sample carrier, a sample collection device, a sample preparation device, an impedance detection device, and a data processing device.

[0042] In one implementation, the data processing device may be independently arranged at a remote end, for example, on another computer. Figure 1 As shown, the blood analysis system includes a blood analyzer 100 and a computer device 200, wherein the blood analyzer 100 includes a sample collection device 110, a sample preparation device 120, and an impedance detection device 130, and the computer device 200 includes a data processing device 150, and the computer device 200 can communicate with the blood analyzer 100. For example, the data processing device 150 can obtain an impedance pulse signal from the blood analyzer 100.

[0043] In one implementation, Figure 2As shown, the blood analysis system is a blood analyzer 100 , which includes a sample collection device 110 , a sample preparation device 120 , an impedance detection device 130 , and a data processing device 150 . Furthermore, illustratively, the blood analyzer 100 may further include a display device 140 .

[0044] In some embodiments, as Figure 2 As shown, the blood analyzer 100 further includes a first housing 160 and a second housing 170. The impedance detection device 130 and the data processing device 150 are disposed within the second housing 170, on either side of the second housing 170. The sample preparation device 120 is disposed within the first housing 160. The display device 140 is disposed on the outer surface of the first housing 160.

[0045] For example, the sample collection device 110 is used to collect a blood sample and transfer the blood sample to the sample preparation device 120. The sample collection device 110 may include a pipette with a pipette nozzle (e.g., a sampling needle) and a drive unit. The drive unit is used to drive the pipette to quantitatively aspirate the blood sample to be tested through the pipette nozzle. For example, the sampling needle is driven by the drive unit to move to aspirate the blood sample to be tested from a sample container containing the blood sample.

[0046] In one example, the sample preparation device 120 is used to process a blood sample using a reagent to prepare a sample solution to be tested. For example, the sample preparation device 120 may include at least one reaction cell and a reagent supply device. The at least one reaction cell is used to receive the blood sample to be tested drawn by the sample collection device, and the reagent supply device provides a processing reagent to the at least one reaction cell. The blood sample to be tested drawn by the sample collection device and the processing reagent provided by the reagent supply device are mixed in the at least one reaction cell to prepare the sample solution to be tested. In some embodiments, the reagent supply device includes a reagent supply portion for supplying a red blood cell reagent, such as a diluent.

[0047] Exemplarily, the sample carrying device is used to carry the sample liquid to be tested. The sample carrying device may include any device capable of carrying the sample liquid to be tested, such as a sample container. Optionally, the sample container includes but is not limited to a test tube, a reaction cup, and the like.

[0048] Illustratively, the impedance detection device may be used to detect the sample liquid to be tested to obtain an impedance pulse signal of cells in the sample liquid to be tested.

[0049] In some embodiments, the impedance detection device 130 includes an impedance detection unit 132 for detecting a sample solution prepared from a portion of the sample solution and a red blood cell reagent supplied from a reagent supply unit to obtain an impedance pulse signal.

[0050] For example, the impedance detection unit 132 is configured as a sheath flow impedance detection unit, such as Figure 3 As shown, the sheath flow impedance detection unit 132 includes a flow chamber 1321 having a well 1322 with an electrode 1323. The sheath flow impedance detection unit 132 detects the DC impedance generated when particles in the sample solution pass through the well 1322 and outputs an impedance pulse signal reflecting the information of the particles passing through the well.

[0051] Specifically, after drawing a blood sample, the sample collection device 110 is driven by its drive device and moved to the reaction pool of the sample preparation device 120, where the drawn blood sample is injected into the reaction pool. A delivery pipeline delivers the sample liquid to be tested, after being treated with a diluent in the reaction pool, to the sheath flow impedance detection unit 132, namely, to the flow chamber 1321. The sheath flow impedance detection unit 132 may also be provided with a sheath liquid chamber (not shown) for supplying sheath liquid to the flow chamber 1321. In the flow chamber 1322, the sample liquid to be tested flows under the sheath liquid. The small holes 1322 transform the sample liquid flow into a thin stream, allowing the particles (formed components) contained in the sample to be tested to pass through the small holes 1322 one by one. The electrodes 1323 are electrically connected to a DC power supply 1324, which provides DC power between the pair of electrodes 1323. While the DC power supply 1324 is providing DC power, the impedance between the pair of electrodes 1323 can be detected. The impedance pulse signal indicating the impedance change is amplified by the amplifier 1325 and then transmitted to the data processing device 150 .

[0052] In one embodiment, the display device 140 is configured to display information related to blood cell parameters. For example, the display device 140 is configured as a user interface.

[0053] In one embodiment, Figure 4As shown, the data processing device 150 includes at least a processing component 151, RAM 152, ROM 153, a communication interface 154, a memory 156, and an I / O interface 155. The processing component 151, RAM 152, ROM 153, the communication interface 154, the memory 156, and the I / O interface 155 communicate via a bus 157. The processing component can be a CPU, a GPU, or other chips with computing capabilities. The memory 156 contains various computer programs, such as an operating system and application programs, for execution by the processor component 151, as well as the data required to execute the computer programs. In addition, during the blood sample analysis process, any data that needs to be stored locally can be stored in the memory 156. The I / O interface 155 is composed 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 155 is connected to an input device such as a keyboard, mouse, touch screen, or other control buttons, allowing the user to directly input data into the data processing device 150. Furthermore, the I / O interface 155 can also be connected to a display device 140 having a display function, such as an LCD screen, a touch screen, or an LED display. The data processing device 150 can output processed data to the display device 140 as graphical display data, such as blood cell parameters and instrument operating parameters. The communication interface 154 can be an interface that uses any currently known communication protocol. The communication interface 154 communicates with the outside world via a network. The data processing device 150 can transmit data to any device connected to the network using a specific communication protocol via the communication interface 154.

[0054] In some embodiments, the blood analyzer 100 further includes a blood sample distribution device (not shown). The blood sample distribution device is configured to divide the sample liquid to be tested, drawn by the sample collection device 110, into at least two different blood sample portions for use in preparing the sample liquid to be tested for different blood cell parameters. For example, the blood sample distribution device is configured to divide the sample liquid to be tested into a first blood sample portion and a second blood sample portion. For example, the first blood sample portion is used to test blood cell parameters such as reticulocyte ratio (RET%) and immature platelet ratio (IPF), while the second blood sample portion is used to test blood cell parameters such as platelet count (PLT). The blood sample distribution device can be configured, for example, as a blood diversion valve. Alternatively, the blood sample distribution device can be a sampling needle of the sample collection device 110. In this case, the driving unit of the sample collection device 110 drives the sampling needle to different reaction cells of the sample preparation device 120 to distribute portions of the sample liquid to be tested to the different reaction cells for reaction with corresponding processing reagents.

[0055] In some embodiments, the data processing device 150 includes a processor and a storage medium storing a computer program. The data processing device is configured to perform the following steps when the computer program is executed by the processor: obtaining a blood cell histogram of the sample liquid to be tested based on the impedance pulse signal, inputting the blood cell histogram into a pre-trained deep learning neural network model, and obtaining blood cell parameters output by the deep learning neural network model, wherein the blood cell parameters are at least one of the following: reticulocyte parameters, reticulocyte platelet parameters, and red blood cell fragment parameters; wherein the deep learning neural network model is obtained by training blood training samples marked with blood cell parameters through a machine learning algorithm.

[0056] In one example, the reticulocyte parameter may include the reticulocyte number (RET#) and / or the reticulocyte fraction (RET%). In another example, the reticulocyte parameter may include the reticulocyte number and / or the reticulocyte fraction (IPF%). In another example, the red blood cell fragment parameter may include the red blood cell fragment number (FRC#) and / or the red blood cell fragment fraction (FRC%).

[0057] Furthermore, the data processing device 150 is further configured to control the display device 140 to display the blood cell parameters of the sample solution to be tested. Furthermore, the data processing device 150 is configured to simultaneously control the display device 140 to display the blood cell parameters of the sample solution to be tested and to display an alarm message. For example, an alarm message may be displayed when any blood cell parameter exceeds a corresponding threshold value.

[0058] In one implementation, the embodiment of the present invention obtains a deep learning neural network model through training. For example, the deep learning neural network model can have a network structure similar to AlexNet, for example, including multiple convolutional layers, pooling layers, fully connected layers (FC) and regression layers.

[0059] like Figure 5 The figure shows an embodiment of this implementation. The blood sample collected by the sample collection device can react with the reagent in the sample preparation device, and the prepared sample liquid to be tested enters the impedance detection device to obtain an impedance pulse signal. Optionally, the impedance pulse signal may include a voltage pulse signal.

[0060] exist Figure 5 In the example, the data processing device can obtain an input impedance pulse signal such as a voltage pulse signal, and obtain a blood cell histogram of the sample liquid to be tested, such as an RBC / PLT statistical histogram, according to the impedance pulse signal. In addition, Figure 5The deep learning neural network model shown includes AlexNet, fully connected layers (FC-4096, FC-1024, FC-2), and regression layers. Figure 5 The outputs of the deep learning neural network model shown in include IPF%, RET%, and FRC%.

[0061] Understandably, Figure 5 The figures shown are for illustration only. This application does not limit the specific structure of the deep learning model, and the output of the deep learning model is not limited to IPF%, RET%, and FRC%, but can be at least one of RET#, RET%, IPF%, FRC#, and FRC%.

[0062] For example, the deep learning neural network model is obtained by training blood training samples with blood cell parameter markers using a machine learning algorithm, wherein the blood training samples may include a large amount of training data.

[0063] In one example, when the blood cell parameters predetermined for output by the deep learning neural network model include reticulocyte parameters, the blood training samples are blood training samples marked with reticulocyte parameters, and the blood training samples include blood training samples with reticulocytes and blood training samples without reticulocytes, wherein the reticulocyte parameter mark corresponding to the blood training samples without reticulocytes is 0.

[0064] In one example, when the blood cell parameters predetermined for output by the deep learning neural network model include reticulated platelet parameters, the blood training samples are blood training samples marked with reticulated platelet parameters, and the blood training samples include blood training samples with reticulated platelets and blood training samples without reticulated platelets, wherein the reticulated platelet parameter corresponding to the blood training samples without reticulated platelets is marked as 0.

[0065] In one example, when the blood cell parameters predetermined for output by the deep learning neural network model include red blood cell fragment parameters, the blood training samples are blood training samples marked with reticulated platelet parameters, and the blood training samples include blood samples with cell fragment parameters and blood samples without cell fragment parameters, the cell fragment parameters corresponding to the blood samples without cell fragment parameters are marked as 0.

[0066] During training, a blood cell histogram is obtained based on the above-mentioned blood training sample. The deep learning neural network model obtains corresponding blood cell parameters by learning the morphological features of the blood cell histogram corresponding to different blood cell parameters. For example, when the blood cell histogram includes red blood cell volume distribution, the deep learning neural network model obtains reticulocyte parameters by learning the morphological features of the red blood cell volume distribution. For another example, when the blood cell histogram includes platelet volume distribution, the deep learning neural network model obtains the reticulocyte parameter by learning the morphological features of the platelet volume distribution. For another example, when the blood cell histogram includes red blood cell volume distribution and platelet volume distribution, the deep learning neural network model obtains red blood cell fragment parameters by learning the morphological features of the transition region between the red blood cell volume distribution and the platelet volume distribution.

[0067] Optionally, the above-mentioned morphological features include but are not limited to: one or more of the following features: pulse width, impedance channel volume statistical histogram tail ratio, pulse peak value, etc., wherein the blood cell parameters of each training data can be obtained based on optical signal detection, and the obtained blood cell parameters are used as the annotation information of the corresponding training data, that is, the annotation information includes at least one of RET#, RET%, IPF%, FRC#, and FRC%, and the blood cell histogram of each training data is obtained based on the impedance detection device. The training data can be input into the deep learning neural network model to be trained, and the deep learning neural network model to be trained learns the morphology of the blood cell histogram corresponding to the different blood cell parameters of each training data, that is, learns the blood cell parameters corresponding to the different morphological histograms, and inputs the blood cell histogram into the deep learning neural network model to be trained to obtain an output (the output is also the blood cell parameter). A loss function is constructed based on the corresponding output and the annotation information, so that it can be judged whether the training is completed based on the loss function.

[0068] Among them, the method of obtaining blood cell parameters based on optical signal detection can refer to the existing optical blood cell analyzer; the method of training the deep learning neural network model can refer to the existing training process such as neural network; it will not be repeated in this application.

[0069] In this way, the present application can obtain blood cell parameters such as RET#, RET%, IPF%, FRC#, FRC%, etc. based on the deep learning neural network model, without the need for fluorescent staining of cells, making the operation simpler and saving detection costs.

[0070] like Figures 6 to 8 A schematic diagram of the correlation between the blood cell parameters obtained by the scheme of the present application and the blood cell parameters obtained by the fluorescent staining scheme is shown. It can be seen that the blood cell parameters obtained by the present application are highly accurate.

[0071] Figure 6 The horizontal axis (x) represents the RET% obtained by fluorescent staining, and the vertical axis (y) represents the RET% obtained by the deep learning neural network model of this application. It can be seen that the correlation between the two satisfies: y = 0.649x + 0.7595, and the correlation coefficient is R 2 =0.6573.

[0072] Figure 7 The horizontal axis (x) represents the IPF% obtained by fluorescent staining, and the vertical axis (y) represents the IPF% obtained by the deep learning neural network model of this application. It can be seen that the correlation between the two satisfies: y = 0.7663x + 1.2235, and the correlation coefficient is R 2 =0.7063.

[0073] Figure 8 The horizontal axis (x) represents the FRC% obtained by fluorescent staining, and the vertical axis (y) represents the FRC% obtained by the deep learning neural network model of this application. It can be seen that the correlation between the two satisfies: y = 0.8422x + 0.176, and the correlation coefficient is R 2 =0.8293.

[0074] In one example, the data processing device 150 is further configured to obtain a blood cell histogram based on the cell impedance pulse signal. The blood cell histogram reflects the volume distribution of various blood cells in the blood sample to be tested, such as the volume distribution of red blood cells, platelets, small red blood cells, and red blood cell fragments. The data processing device 150 is further configured to input the blood cell histogram into a pre-trained deep learning neural network model to obtain an interference-removed blood cell histogram output by the deep learning neural network model. The training samples used to train the deep learning neural network model specifically include blood cell histogram samples and interference-removed blood cell histograms corresponding to the blood cell histogram samples.

[0075] Optionally, the blood cell histogram without interference removal reflects the platelet volume distribution, small-volume red blood cell volume distribution and normal red blood cell volume distribution in the sample liquid to be tested, wherein the small-volume red blood cell volume distribution and the platelet body distribution overlap, thereby interfering with platelet counting, etc.

[0076] Optionally, the blood cell histogram output by the deep learning neural network model is a histogram that excludes the volume distribution of small-volume red blood cells.

[0077] Specifically, the deep learning neural network model training data set obtained by pre-training includes a blood cell histogram with interference (such as an impedance channel volume histogram) and a corresponding undisturbed blood cell histogram obtained based on optical signal detection, such as an optical channel volume histogram. The deep learning neural network model learns the impedance channel volume histogram of the interfered blood sample and the undisturbed optical channel volume histogram obtained by optical PLT-O detection. By learning the above two types of volume histograms, the feature is mined, which is used for regression prediction to remove interference, and the interference of small-volume red blood cells and red blood cell fragments in the impedance channel statistical histogram is removed to obtain an output. The output can be a blood cell histogram after the interference is removed, and the blood cell histogram after the interference is removed can be used to count platelets, etc. The method for training the deep learning neural network model can refer to the existing training process such as neural network; it will not be repeated in this application.

[0078] like Figure 9 The figure shows an embodiment of this implementation. The blood sample collected by the sample collection device can react with the reagent in the sample preparation device, and the prepared sample liquid to be tested enters the impedance detection device to obtain an impedance pulse signal such as a voltage pulse signal.

[0079] exist Figure 9 In the example, the data processing device 150 can be used to obtain a blood cell histogram, such as an RBC / PLT volume statistical histogram, based on the impedance pulse signal of the cell. The blood cell histogram reflects the platelet volume distribution, small-volume red blood cell volume distribution and normal red blood cell volume distribution in the sample liquid to be tested, wherein the small-volume red blood cell volume distribution and the platelet body distribution overlap.

[0080] In order to eliminate the interference information in the initial blood cell histogram, a deep learning neural network model can be pre-trained. The trained deep learning neural network model has the following prediction capabilities: it uses deep learning methods to mine the channel information of the impedance detection channel and predicts the platelet histogram after removing the interference information through regression. A structure of the deep learning neural network model is as follows: Figure 9As shown, it includes an Alex-Net layer, FC-4096, FC-4096, FC-256, and a regression layer. The data processing device inputs the initial blood cell histogram into the trained deep learning neural network model, which can remove the interference of red blood cell pulse signals with a volume less than the volume threshold on the platelet statistical results from the initial blood cell histogram, thereby obtaining a blood cell histogram containing only the platelet volume distribution (that is, a blood cell histogram with interference removed), thereby improving the accuracy of the platelet statistical results. In actual applications, various analysis and processing of platelets can be performed based on the blood cell histogram with interference removed according to processing requirements, including but not limited to platelet counting.

[0081] The data processing device 150 can also obtain platelet parameters based on the interference-free blood cell histogram, including the reticulated platelet ratio and / or platelet count (PLT#). Alternatively, the deep learning neural network model can directly obtain and output platelet parameters using the interference-free blood cell histogram instead of outputting the interference-free blood cell histogram.

[0082] Understandably, Figure 9 The illustration is for illustrative purposes only, and this application does not limit the specific structure of the deep learning neural network model.

[0083] In existing hematology analyzers, platelet counts (PLT#) and red blood cell counts (RBC#) are performed in the same channel. For normal samples, due to the volume difference between PLT and RBC, the two particles can be distinguished by volume information, and they do not interfere with each other during counting. Figure 10 As shown in the figure, the volume larger than the volume dividing line is RBC, and the volume smaller than the volume dividing line is PLT. However, when there are a large number of small red blood cells or red blood cell fragments in the blood sample, since the number of RBC is usually much larger than PLT, these small red blood cells and red blood cell fragments will seriously affect the platelet count, resulting in a falsely high platelet count value, such as Figure 11 As shown, small-volume red blood cells and red blood cell fragments larger than those near the volume cutoff line may be regarded as PLTs, causing the PLT# to be too large.

[0084] In this application, a deep learning neural network model is used to predict the blood cell histogram after removing the interference of red blood cell fragments and small-volume red blood cells through regression. Figure 12, where the PLT obtained by the prior art is an initial blood cell histogram directly obtained based on the volume size distribution on the basis of impedance channel information (i.e., the impedance channel PLT statistical histogram), and the PLT obtained in this application is a blood cell histogram with interference removed obtained through a deep learning neural network model (i.e., the PLT statistical histogram predicted by using the deep learning neural network model). The abscissa of the histogram represents the volume of blood cells, and the dotted line represents the boundary with a predetermined volume value. The left and right sides of the initial blood cell histogram and the blood cell histogram with interference removed are the statistical results of platelets and red blood cells respectively. By comparing the statistical results of platelets on the left side of the dotted line in the two histograms, it can be seen that due to the influence of small-volume red blood cells and red blood cell fragments, the statistical result of platelets in the initial blood cell histogram is higher than the statistical result of platelets in the blood cell histogram with interference removed, as Figure 12 shown by the oblique shaded area in

[0085] In one embodiment, the display device 140 can be used to display information related to blood cell parameters. As an example, by using a pre-trained deep learning neural network model to obtain reticulocyte platelet parameters such as the number of reticulocyte platelets and / or the ratio of reticulocyte platelets, the blood cell histogram with interference removed (also referred to as the PLT histogram or the volume statistical histogram of PLT) can be further displayed by the display device 140, as Figure 12 shown in

[0086] In one embodiment, the display device 140 can also be used to display alarm prompt information.

[0087] In one example, if the blood cell parameters obtained by the data processing device 150 include RET#, and it is outside the threshold range of RET (such as [RET1#, RET2#]), for example, RET# < RET1# or RET# > RET2#, then the display device 140 can display the first alarm prompt information. For example, the first alarm prompt information includes RET# and [RET1#, RET2#], and can also include whether the value exceeding the threshold range is too large or too small, and the RET# can also be displayed in a specific display manner (such as the first color, etc.).

[0088] In another example, if the blood cell parameters obtained by the data processing device 150 include FRC%, and it is outside the threshold range of FRC (such as [FRC1%, FRC2%]), for example, FRC% < FRC1% or FRC% > FRC2%, then the display device 140 can display a second alarm prompt message. For example, the second alarm prompt message includes FRC% and [FRC1%, FRC2%], and can also include whether the excess of the threshold range is too large or too small, and can also display the FRC% in a specific display manner (such as a second color, etc.).

[0089] Similarly, for other blood cell parameters, such as RET%, IPF%, FRC#, PLT, when they are outside their respective corresponding threshold ranges, the corresponding alarm prompt messages can be displayed similarly, which will not be listed one by one here.

[0090] An embodiment of the present invention also provides a blood analysis method, as Figure 13 shown, the method includes the following steps:

[0091] S10, obtaining impedance pulse signals of cells in the test sample solution detected by an impedance detection device for the test sample solution.

[0092] S20, obtaining a blood cell histogram of the test sample solution according to the impedance pulse signals.

[0093] S30, inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, where the blood cell parameters are at least one of the following: reticulocyte parameters, reticulocyte platelet parameters, and red blood cell fragment parameters; wherein, the deep learning neural network model is obtained by training a blood training sample with blood cell parameter labels through a machine learning algorithm.

[0094] Exemplarily, before S10, it may include: collecting a blood sample by a sample collection device of the blood analysis system; mixing the blood sample with a reagent in at least one reaction pool by a sample preparation device of the blood analysis system to prepare a test sample solution.

[0095] In step S30, the training method of the deep learning neural network model can refer to the previous description and will not be described one by one here.

[0096] An embodiment of the present invention also provides a blood analysis method, as Figure 14 shown, the method includes the following steps:

[0097] S110, obtaining impedance pulse signals of cells in the test sample solution detected by an impedance detection device for the test sample solution;

[0098] S120, obtaining a blood cell histogram of the sample liquid to be tested according to the impedance pulse signal;

[0099] S130: Inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain an interference-removed blood cell histogram output by the deep learning neural network model. The training samples used to train the deep learning neural network model specifically include blood cell histogram samples and interference-removed blood cell histograms corresponding to the blood cell histogram samples.

[0100] Optionally, a blood cell histogram, such as an impedance channel volume histogram, reflects the platelet volume distribution, small-volume red blood cell volume distribution and normal red blood cell volume distribution in the sample liquid to be tested, wherein the small-volume red blood cell volume distribution and the platelet body distribution overlap; the blood cell histogram output by the pre-trained deep learning neural network model is a histogram excluding the small-volume red blood cell volume distribution, which may include the platelet volume distribution, or may also include the platelet volume distribution and the red blood cell volume distribution.

[0101] Understandable, such as Figure 13 or Figure 14 The method shown can be executed by the above-mentioned data processing device 150, and specifically can be executed by the processor included in the data processing device 150.

[0102] The embodiment of the present invention further provides a blood analysis device, which includes a processor and a memory, wherein the memory stores a computer program and the computer program can be run on the processor to perform the above-mentioned Figure 13 The method shown or the execution of the above Figure 14 The method shown.

[0103] For example, the blood analysis device may be the aforementioned data processing device 150. As an example, the blood analysis device may be Figure 4 shown.

[0104] In addition, an embodiment of the present invention further provides a computer storage medium on which a computer program is stored. When the computer program is executed by a computer or a processor, the aforementioned combination Figure 13 The steps of the method or the combination thereof Figure 14 For example, the computer storage medium is a computer-readable storage medium.

[0105] In one embodiment, when the computer program instructions are executed by a computer or processor, the computer or processor performs the following steps: obtaining a blood cell histogram of the sample liquid to be tested based on the impedance pulse signal, inputting the blood cell histogram into a pre-trained deep learning neural network model, and obtaining blood cell parameters output by the deep learning neural network model, wherein the blood cell parameters are at least one of the following: reticulocyte parameter, reticulocyte platelet parameter, and red blood cell fragment parameter; wherein the deep learning neural network model is obtained by training blood training samples marked with blood cell parameters using a machine learning algorithm.

[0106] In another embodiment, the computer program instructions, when executed by a computer or processor, cause the computer or processor to perform the following steps: obtaining a blood cell histogram of the sample liquid to be tested based on the impedance pulse signal; and inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain an interference-eliminated blood cell histogram output by the deep learning neural network model. The training samples used to train the deep learning neural network model specifically include blood cell histogram samples and interference-eliminated blood cell histograms corresponding to the blood cell histogram samples.

[0107] Computer storage media may include, for example, a memory card of a smartphone, a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disk read-only memory (CD-ROM), a USB memory, or any combination thereof. A computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0108] In addition, an embodiment of the present invention further provides a computer program product, which includes instructions, and when the instructions are executed by a computer, the computer executes the above-mentioned combination Figure 13 The steps of the method are described.

[0109] It can be seen that in the embodiment of the present invention, the blood cell parameters of the sample liquid to be tested can be obtained based on the pre-trained deep learning neural network model, without the need for fluorescent staining of the cells, the operation is simpler, the detection cost is saved, and the obtained blood cell parameters are highly accurate.

[0110] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely illustrative and are not intended to limit the scope of the present invention. Various changes and modifications may be made therein by those skilled in the art without departing from the scope and spirit of the present invention. All such changes and modifications are intended to be included within the scope of the present invention as claimed in the appended claims.

[0111] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0112] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units described is merely a logical function division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another device, or ignoring or not performing some features.

[0113] In the description provided herein, numerous specific details are described. However, it is understood that embodiments of the present invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0114] Similarly, it should be understood that in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the description of exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this approach to the present invention should not be interpreted as reflecting the intention that the claimed invention requires more features than those explicitly recited in each claim. More precisely, as reflected in the corresponding claims, the inventive point is that the corresponding technical problem can be solved with fewer features than all the features of a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into the detailed description, with each claim itself serving as a separate embodiment of the present invention.

[0115] It will be understood by those skilled in the art that, except where mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or apparatus disclosed herein may be combined in any combination. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature providing the same, equivalent, or similar purpose.

[0116] Furthermore, those skilled in the art will appreciate that although some embodiments described herein include certain features included in other embodiments but not other features, combinations of features from different embodiments are intended to be within the scope of the present invention and to form different embodiments. For example, in the claims, any of the claimed embodiments may be used in any combination.

[0117] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that in practice, a microprocessor or a digital signal processor (DSP) may be used to implement some or all of the functions of some modules in a blood analysis system according to an embodiment of the present invention. The present invention may also be implemented as a device program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0118] It should be noted that the above embodiments illustrate rather than limit the invention, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising several different elements and by means of appropriately programmed computers. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0119] The foregoing description is merely a specific embodiment of the present invention or an illustration of a specific embodiment. The scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be readily conceived by a person skilled in the art within the technical scope disclosed in the present invention are intended to be encompassed by the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A blood analysis system, characterized in that: include: A sample collection device, used to collect a blood sample and transport the blood sample to a sample preparation device; A sample preparation device, used to process the blood sample using a reagent to prepare a sample liquid to be tested; An impedance detection device, configured to detect the sample liquid to be tested to obtain an impedance pulse signal of cells in the sample liquid to be tested; A data processing device is used to perform the following steps: Obtaining a blood cell histogram of the sample liquid to be tested according to the impedance pulse signal, inputting the blood cell histogram into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, wherein the blood cell parameters are at least one of the following: a reticulocyte parameter, a reticulocyte platelet parameter, and a red blood cell fragment parameter; Wherein, the deep learning neural network model is obtained by training blood training samples with blood cell parameter markers through a machine learning algorithm; The blood cell histogram includes blood cell volume distribution, and the deep learning neural network model obtains the blood cell parameters by learning the morphological characteristics of the blood cell volume distribution.

2. The blood analysis system according to claim 1, wherein: The blood cell parameters output by the deep learning neural network model include reticulocyte parameters, and the blood training samples are blood training samples marked with reticulocyte parameters. The blood training samples include blood training samples containing reticulocytes and blood training samples not containing reticulocytes, wherein the reticulocyte parameter corresponding to the blood training samples not containing reticulocytes is marked as 0.

3. The blood analysis system according to claim 1, wherein: The blood cell parameters output by the deep learning neural network model include reticulated platelet parameters, and the blood training samples are blood training samples marked with reticulated platelet parameters. The blood training samples include blood training samples containing reticulated platelets and blood training samples not containing reticulated platelets, wherein the reticulated platelet parameter corresponding to the blood training samples not containing reticulated platelets is marked as 0.

4. The blood analysis system according to any one of claims 1 to 3, characterized in that: The blood cell parameters output by the deep learning neural network model include red blood cell fragment parameters. The blood training samples are blood training samples marked with reticulated platelet parameters. The blood training samples include blood samples containing cell fragment parameters and blood samples not containing cell fragment parameters. The cell fragment parameters corresponding to the blood samples not containing cell fragment parameters are marked as 0.

5. The blood analysis system according to claim 2, wherein: The blood cell histogram includes red blood cell volume distribution, and the deep learning neural network model obtains the reticulocyte parameters by learning the morphological characteristics of the red blood cell volume distribution.

6. The blood analysis system according to claim 3, wherein: The blood cell histogram includes platelet volume distribution, and the deep learning neural network model obtains the reticulated platelet parameters by learning the morphological characteristics of the platelet volume distribution.

7. The blood analysis system according to claim 4, characterized in that The blood cell histogram includes red blood cell volume distribution and platelet volume distribution, and the deep learning neural network model obtains red blood cell fragment parameters by learning the morphological characteristics of the transition area between the red blood cell volume distribution and the platelet volume distribution.

8. The blood analysis system according to claim 1, wherein: Also includes: A display device is used to display the blood cell parameters.

9. The blood analysis system according to claim 1, wherein: Also includes: The display device is used to display an alarm prompt message when any of the blood cell parameters exceeds the corresponding threshold range.

10. The blood analysis system according to claim 1, wherein: The reticulocyte parameter includes the reticulocyte number and / or the reticulocyte ratio, the red blood cell fragment parameter includes the red blood cell fragment number and / or the red blood cell fragment ratio, and the reticulocyte platelet parameter includes the reticulocyte platelet number and / or the reticulocyte platelet ratio.

11. A blood analysis method, characterized in that: include: Acquiring an impedance pulse signal of cells in the sample liquid to be tested obtained by an impedance detection device through detecting the sample liquid to be tested; obtaining a blood cell histogram of the sample liquid to be tested according to the impedance pulse signal; The blood cell histogram is input into a pre-trained deep learning neural network model to obtain blood cell parameters output by the deep learning neural network model, where the blood cell parameters are at least one of the following: Reticulocyte parameters, reticulocyte parameters and red blood cell fragment parameters; Wherein, the deep learning neural network model is obtained by training blood training samples with blood cell parameter markers through a machine learning algorithm; The blood cell histogram includes blood cell volume distribution, and the deep learning neural network model obtains the blood cell parameters by learning the morphological characteristics of the blood cell volume distribution.

12. A blood analysis device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to claim 11 are implemented.

13. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a computer or a processor, the steps of the method according to claim 11 are implemented.

Images