Analysis method and analysis device

Abstract

To provide a novel method for analysis and a novel analyzer for analyzing a cell that has features of cell species.SOLUTION: A method for analyzing a sample including cells includes the steps of: irradiating a measurement sample prepared from the sample with light and detecting light generated from the cell (S12); acquiring feature data of each of cells in the sample on the basis of the detected light (S12); analyzing the feature data by using an artificial intelligence algorithm to classify the cells into a plurality of cell specifies (S13); and generating result data including the result of classification of the cells into a plurality of cell species (S3).SELECTED DRAWING: Figure 10

Description

[Technical Field]

[0001] The present invention relates to an analytical method and an analytical device for analyzing cells in a specimen. [Background technology]

[0002] Patent Document 1 describes a method for classifying white blood cells into subpopulations using fluorescence and light scattering parameters obtained by flowing blood cells through a flow cell. This method classifies white blood cell subpopulations based on multiple parameters, including axial light loss (ALL), intermediate angle forward scatter (IAS), fluorescence (FL1), and polarized side scatter (PSS), which are obtained by irradiating light onto the blood cells flowing through the flow cell. The classified white blood cell subpopulations are displayed on a scatter plot. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2016-514267 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional cell classification methods such as that disclosed in Patent Document 1 alternatively classify a single cell into one type. However, even the same neutrophils may have different morphological characteristics depending on their maturity, and some neutrophils may have morphological characteristics similar to immature granulocytes. Conventional cell classification methods alternatively classify such cells into one type, even if they contain cells with characteristics of normal cells such as neutrophils and abnormal cells such as immature granulocytes.

[0005] An object of the present invention is to provide a new analytical method and apparatus for analyzing cells having characteristics of multiple cell types. [Means for solving the problem]

[0006] The analytical method of the present invention is a method for analyzing a specimen containing cells, comprising: irradiating a measurement sample prepared from the specimen with light to detect light emitted from the cells; and acquiring characteristic data of each of a plurality of cells contained in the specimen based on the detected light; Deep Learning The characteristic data is analyzed using an algorithm to identify each of the cells. The probability that each of the cells corresponds to each of a plurality of cell types and a first cell type and a second cell type are determined based on the probability, and the results are associated with the ID of the specimen, and (1) a first analysis result in which the plurality of cells are counted and / or classified based on the first cell type, and (2) a second analysis result based on the second cell type. The method is characterized in that it generates result data including:

[0007] The analytical device of the present invention is an analytical device for analyzing a specimen containing cells, and includes a detection unit that irradiates a measurement sample prepared from the specimen with light and detects light emitted from the cells, and a signal processing unit that acquires characteristic data of each of a plurality of cells contained in the specimen based on the detected light. , pu an information processing unit having a processor; The processor analyzes the feature data using a deep learning algorithm to determine, for each of the cells, a probability that the cell corresponds to each of a plurality of cell types, and a first cell type and a second cell type based on the probability, and generates result data including (1) a first analysis result in which the plurality of cells are counted and / or classified based on the first cell type and (2) a second analysis result based on the second cell type, in association with the ID of the specimen. [Effects of the Invention]

[0008] According to the present invention, cells that have conventionally been classified into one alternative type can be classified into a plurality of cell types. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing a cell analyzer and a host computer. [Figure 2] FIG. 2 is a block diagram of the measurement unit. [Figure 3] FIG. 3 is a block diagram of the sample preparation unit. [Figure 4] FIG. 4 is a schematic diagram of the FCM detection unit. [Figure 5] FIG. 5 is a diagram illustrating the process of generating a digital signal by the A / D conversion unit. [Figure 6] FIG. 6 is a diagram illustrating the structure of the feature parameter data. [Figure 7] FIG. 7 is a schematic diagram illustrating the process of cell classification based on a deep learning algorithm. [Figure 8]Figure 8 shows the types of cells analyzed by the deep learning algorithm. [Figure 9] FIG. 9 is a block diagram of the processing unit. [Figure 10] FIG. 10 is a flowchart illustrating the operation of the cell analyzer. [Figure 11] FIG. 11 is a flowchart illustrating the test result data generation process. [Figure 12] FIG. 12 is a diagram schematically illustrating a first example of the data structure of the test result data. [Figure 13] FIG. 13 is a diagram schematically illustrating a second example of the data structure of the test result data. [Figure 14] FIG. 14 is a diagram schematically illustrating a third example of the data structure of the test result data. [Figure 15] FIG. 15 is a diagram schematically illustrating a fourth example of the data structure of the test result data. [Figure 16] FIG. 16 is a diagram schematically illustrating a fifth example of the data structure of the test result data. [Figure 17] FIG. 17 is a flowchart of a first example of a test result display process. [Figure 18] FIG. 18 is a diagram showing an example of the sample list screen. [Figure 19] FIG. 19 is a diagram showing an example of the details screen. [Figure 20] FIG. 20 is a diagram showing an example of the analysis screen. [Figure 21] FIG. 21 is a diagram showing an example of the analysis screen. [Figure 22] FIG. 22 is a diagram showing an example of the analysis screen. [Figure 23] FIG. 23 is a diagram showing an example of the analysis screen. [Figure 24] FIG. 24 is a flowchart of a second example of the test result display process. [Figure 25] FIG. 25 is a diagram showing an example of the analysis screen. [Figure 26] FIG. 26 is a diagram showing an example of the analysis screen. [Figure 27] FIG. 27 is a diagram showing an example of the analysis screen. [Figure 28] FIG. 28 is a diagram showing an example of the analysis screen. [Figure 29] FIG. 29 is a flowchart of a third example of the test result display process. [Figure 30] FIG. 30 is a flowchart of a fourth example of the test result display process. [Figure 31] FIG. 31 is a diagram showing an example of the analysis screen. [Figure 32] FIG. 32 is a diagram showing an example of the analysis screen. [Figure 33] FIG. 33 is a diagram showing an example of the analysis screen. [Figure 34] FIG. 34 is a diagram showing an example of the analysis screen. [Figure 35] FIG. 35 is a diagram showing an example of the analysis screen. [Figure 36] FIG. 36 is a diagram showing an example of the analysis screen. [Figure 37] FIG. 37 is a diagram showing an example of the analysis screen. [Figure 38] FIG. 38 is a diagram showing an example of the analysis screen. [Figure 39] 39(a) and 39(b) are diagrams showing other examples of scattergrams. [Figure 40] FIG. 40 is a flowchart of a fifth example of the test result display process. [Figure 41] FIG. 41 is a diagram showing an example of the analysis screen. [Figure 42] FIG. 42 is a diagram showing an example of the analysis screen. [Figure 43] FIG. 43 is a flow chart of the transmission process to the host computer. [Figure 44] 44(a) and 44(b) are diagrams showing examples of the structure of output data. [Figure 45] 45(a) and 45(b) are diagrams showing examples of the structure of output data. [Figure 46]46(a) and 46(b) are diagrams showing examples of screens for accepting settings related to output data. [Figure 47] FIG. 47 is a diagram showing an example of the configuration of a parallel processing processor. [Figure 48] FIG. 48 is a diagram showing an example of mounting a parallel processing processor on a measurement unit. [Figure 49] FIG. 49 is a diagram showing an example of mounting a parallel processing processor on a measurement unit. [Figure 50] FIG. 50 is a diagram showing an example of mounting a parallel processing processor on a measurement unit. [Figure 51] FIG. 51 is a diagram showing an example of mounting a parallel processing processor on a measurement unit. [Figure 52] FIG. 52 is a diagram showing an outline of the arithmetic processing executed by the processor and the parallel processing processor. [Figure 53] 53(a) and 53(b) are diagrams each showing an outline of a matrix operation executed by a parallel processor. [Figure 54] FIG. 54 is a diagram showing that a plurality of arithmetic processes are executed by a parallel arithmetic processor. [Figure 55] 55(a) and 55(b) are diagrams each showing an outline of the calculation process related to the convolution layer. [Figure 56] Figure 56 is a flowchart of cell classification using a deep learning algorithm with a processor and a parallel processing processor. [Figure 57] FIG. 57 is a flowchart of parallel processing execution. [Figure 58] FIG. 58 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 59] FIG. 59 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 60] FIG. 60 is a schematic diagram showing the control of the parallel processing processor by the processor in another example of the configuration of the cell analyzer. [Figure 61] FIG. 61 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 62] FIG. 62 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 63] FIG. 63 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 64] FIG. 64 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 65] FIG. 65 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 66] FIG. 66 is a schematic diagram showing the control of a parallel processing processor by a processor in another example of the configuration of the cell analyzer. [Figure 67] FIG. 67 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 68] FIG. 68 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 69] FIG. 69 is a block diagram showing another example of the configuration of the cell analyzer. [Figure 70] 70(a) to (c) are diagrams each explaining the structure of a neural network. [Figure 71] FIG. 71 is a diagram illustrating the training of a neural network. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an overview and embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description and drawings, the same reference numerals will denote the same or similar components, and descriptions of the same or similar components will be omitted.

[0011] The analytical method described below involves irradiating a measurement sample prepared from a specimen containing cells with light to detect the light emitted from the cells, obtaining cell characteristic data for each of the multiple cells contained in the specimen based on the detected light, and analyzing the characteristic data using an artificial intelligence algorithm to classify each cell into multiple cell types.

[0012] The specimen may be a biological sample collected from a subject. For example, the biological sample may include peripheral blood such as venous blood and arterial blood, urine, and body fluids other than blood and urine. Body fluids other than blood and urine may include bone marrow fluid, ascites, pleural effusion, cerebrospinal fluid, etc. Hereinafter, body fluids other than blood and urine may be simply referred to as "body fluids." A blood sample is one that allows cell counting and cell type determination, i.e., one that contains cells, and is preferably whole blood. The blood is preferably peripheral blood. For example, the blood may be peripheral blood collected using an anticoagulant such as ethylenediaminetetraacetic acid (sodium salt or potassium salt) or heparin sodium. Peripheral blood may be collected from an artery or a vein.

[0013] The cell types classified by this analysis method are based on cell types based on morphological classification and vary depending on the type of biological sample. When the biological sample is blood and the blood is collected from a healthy individual, the cell types to be determined in this embodiment include, for example, nucleated cells such as red blood cells and white blood cells, platelets, etc. Nucleated cells include, for example, neutrophils, lymphocytes, monocytes, eosinophils, and basophils. Neutrophils include, for example, segmented neutrophils and band-shaped neutrophils. When the blood is collected from a non-healthy individual, the nucleated cells include abnormal cells.

[0014] Abnormal cells refer to cells that are not normally found in the peripheral blood of healthy individuals. Abnormal cells may include abnormal lymphoid cells. Abnormal lymphoid cells include, for example, atypical lymphocytes (reactive lymphocytes), abnormal lymphocytes including mature lymphomas, and plasma cells. Abnormal cells may include blasts. Blasts include myeloblasts, lymphoblasts, proerythroblasts, basophilic erythroblasts, polychromatic erythroblasts, normochromatic erythroblasts, promegaloblasts, basophilic megaloblasts, polychromatic megaloblasts, and normochromatic megaloblasts. Abnormal cells may include megakaryocytes. Abnormal cells may include immature granulocytes. Immature granulocytes include, for example, promyelocytes, myelocytes, and metamyelocytes.

[0015] Abnormal cells may include other abnormal cells not found in the peripheral blood of healthy individuals. Examples of abnormal cells are cells that appear when a patient is affected with a disease, such as tumor cells. In the case of the hematopoietic system, examples of the disease include leukemias such as myelodysplastic syndrome, acute myeloblastic leukemia, acute myeloblastic leukemia, acute promyelocytic leukemia, acute myelomonocytic leukemia, acute monocytic leukemia, erythroleukemia, acute megakaryoblastic leukemia, acute myeloid leukemia, acute lymphocytic leukemia, lymphoblastic leukemia, chronic myelogenous leukemia, and chronic lymphocytic leukemia, malignant lymphomas such as Hodgkin's lymphoma and non-Hodgkin's lymphoma, and multiple myeloma.

[0016] When the biological sample is urine, the cell types may include, for example, red blood cells, white blood cells, epithelial cells such as transitional epithelium and squamous epithelium, etc. The abnormal cells may include, for example, bacteria, fungi such as filamentous fungi and yeast, tumor cells, etc.

[0017] When the biological sample is a body fluid that does not normally contain blood components, such as ascites, pleural effusion, or cerebrospinal fluid, the cell types may include, for example, red blood cells, white blood cells, and large cells. Here, "large cells" refer to cells that are detached from the lining of a body cavity or the peritoneum of an internal organ and are larger than white blood cells, such as mesothelial cells, histiocytes, and tumor cells.

[0018] When the biological sample is bone marrow fluid, the cell types may include, as normal cells, mature blood cells and immature blood cells. Mature blood cells include, for example, nucleated cells such as red blood cells and white blood cells, and platelets. Nucleated cells such as white blood cells include, for example, neutrophils, lymphocytes, plasma cells, monocytes, eosinophils, and basophils. Neutrophils include, for example, segmented neutrophils and band-shaped neutrophils. Immature blood cells include, for example, hematopoietic stem cells, immature granulocytic cells, immature lymphocytic cells, immature monocytic cells, immature erythroid cells, megakaryocytic cells, mesenchymal cells, and the like. Immature granulocytes may include, for example, metamyelocytes, myelocytes, promyelocytes, myeloblasts, and the like. Immature lymphocytic cells include, for example, lymphoblasts, and the like. Immature monocytic cells include, for example, monoblasts, and the like. Immature erythroid cells include nucleated erythrocytes such as proerythroblasts, basophilic erythroblasts, polychromatic erythroblasts, normochromatic erythroblasts, promegaloblasts, basophilic megaloblasts, polychromatic megaloblasts, and normochromatic megaloblasts. Megakaryocytic cells include megakaryoblasts, etc.

[0019] Abnormal cells that may be contained in bone marrow include, for example, leukemias such as the above-mentioned myelodysplastic syndrome, acute myeloblastic leukemia, acute myeloblastic leukemia, acute promyelocytic leukemia, acute myelomonocytic leukemia, acute monocytic leukemia, erythroleukemia, acute megakaryoblastic leukemia, acute myeloid leukemia, acute lymphocytic leukemia, lymphoblastic leukemia, chronic myeloid leukemia, and chronic lymphocytic leukemia; hematopoietic tumor cells of malignant lymphomas such as Hodgkin's lymphoma and non-Hodgkin's lymphoma; and metastatic tumor cells of malignant tumors that have developed in organs other than the bone marrow.

[0020] This analysis method is preferably carried out using a cell analyzer that analyzes a specimen containing cells. The cell analyzer may include a detection unit that irradiates a measurement sample prepared from the specimen with light and detects light emitted from the cells, a signal processing unit that acquires cell characteristic data for each of a plurality of cells contained in the specimen based on the detected light, and a control unit that analyzes the characteristic data using an artificial intelligence algorithm to classify each cell into a plurality of cell types. An example of such a cell analyzer is described below.

[0021] [1.Basic configuration] The basic configuration of a cell analyzer will be described with reference to Fig. 1. Fig. 1 is a schematic diagram showing the appearance of a cell analyzer 100. The cell analyzer 100 is an apparatus that analyzes biological samples in accordance with a test order transmitted from a host computer 500. The cell analyzer 100 comprises a measurement unit 400 and a processing unit 300. The host computer 500 and the cell analyzer 100 are collectively referred to as a test system 1000.

[0022] [2. Measuring unit configuration] The configuration of the measurement unit 400 will be described with reference to Fig. 2. Fig. 2 shows an example of a block diagram of the measurement unit 400. As shown in Fig. 2, the measurement unit 400 includes a specimen aspirating section 450 that aspirates a specimen, a specimen preparing section 440 that prepares a measurement specimen from the aspirated specimen, an FCM detection section 410 that detects blood cells in the measurement specimen, an analog processing section 420 that processes analog signals output from the FCM detection section 410, a measurement unit control section 480 that converts signals processed by the analog processing section 420 into digital signals for analysis, and an apparatus mechanism section 430.

[0023] 3 is a schematic diagram illustrating the specimen aspirating unit 450 and the sample preparing unit 440. The specimen aspirating unit 450 includes a nozzle 451 for aspirating a blood specimen from a blood collection tube T, and a pump 452 for applying negative / positive pressure to the nozzle 451. The nozzle 451 is inserted into the blood collection tube T by being moved up and down by the device mechanism unit 430. When the pump 452 applies negative pressure while the nozzle 451 is inserted into the blood collection tube T, the blood specimen is aspirated through the nozzle 451.

[0024] The sample preparation unit 440 includes five reaction chambers 440a to 440e. The reaction chambers 440a to 440e are used in the DIFF, RET, WPC, PLT-F, and WNR measurement channels, respectively. A hemolytic agent container containing a hemolytic agent and a staining solution container containing a staining solution, which are reagents corresponding to the measurement channel, are connected to each reaction chamber via a flow path. A measurement channel is formed by one reaction chamber and the reagents (hemolytic agent and staining solution) connected thereto. For example, the DIFF measurement channel is formed by the DIFF hemolytic agent and DIFF staining solution, which are DIFF measurement reagents, and reaction chamber 440a. The other measurement channels are similarly configured. Note that, although the configuration in which one measurement channel contains one hemolytic agent and one staining solution is illustrated here, one measurement channel does not necessarily have to contain both a hemolytic agent and a staining solution; one reagent may be shared by multiple measurement channels.

[0025] After aspirating the blood sample, nozzle 451 accesses one of reaction chambers 440a-440e corresponding to the measurement channel corresponding to the order from above by horizontal and vertical movement using device mechanism unit 430, and dispenses the aspirated blood sample. Sample preparation unit 440 supplies the corresponding hemolytic agent and staining solution to the reaction chamber into which the blood sample has been dispensed, and prepares a measurement sample by mixing the blood sample, hemolytic agent, and staining solution in the reaction chamber. The prepared measurement sample is supplied from the reaction chamber via a flow path to FCM detection unit 410, where cells are measured by flow cytometry.

[0026] Fig. 4 shows an example of the configuration of the optical system of the FCM detection unit 410. As shown in Fig. 4, in measurement using a flow cytometer, when cells contained in a measurement sample pass through a flow cell (sheath flow cell) 4113 provided in the flow cytometer, a light source 4111 irradiates the flow cell 4113 with light, and scattered light and fluorescent light emitted from the cells in the flow cell 4113 by this light are detected by light receiving elements 4116, 4121, and 4122.

[0027] In FIG. 4, light emitted from a laser diode, which is a light source 4111, is irradiated onto cells passing through a flow cell 4113 via an illumination lens system 4112.

[0028] As shown in FIG. 4 , forward scattered light emitted from particles passing through a flow cell 4113 is received by a light receiving element 4116 via a condenser lens 4114 and a pinhole 4115. The light receiving element 4116 is, for example, a photodiode. The side scattered light is received by a light receiving element 4121 via a condenser lens 4117, a dichroic mirror 4118, a bandpass filter 4119, and a pinhole 4120. The light receiving element 4121 is, for example, a photomultiplier tube. The side fluorescent light is received by a light receiving element 4122 via the condenser lens 4117 and a dichroic mirror 4118. The light receiving element 4122 is, for example, a photomultiplier tube. Note that an avalanche photodiode or a photomultiplier tube may be used as the light receiving element 4116 instead of a photodiode. The light receiving elements 4121 and 4122 may be photodiodes or avalanche photodiodes.

[0029] The light receiving signals output from the light receiving elements 4116, 4121, and 4122 are input to an analog processing unit 420 via amplifiers 4151, 4152, and 4153, respectively. The analog processing unit 420 is connected via a signal transmission path 421 to an A / D conversion unit 482 of a measurement unit control unit 480, which will be described later.

[0030] Returning to FIG. 2, the analog processing section 420 performs processing such as noise removal and smoothing on the analog signal input from the FCM detection section 410, and outputs the processed analog signal to the measurement unit control section 480.

[0031] The measuring unit control section 480 includes an A / D conversion section 482, a processor 4831, a RAM 4834, a storage section 4835, a bus controller 4850, a parallel processing processor 4833, a bus 485, and an interface section 489 that connects to the processing unit 300. The measuring unit control section 480 includes an interface section 484 that is interposed between the bus 485 and the A / D conversion section 482. The measuring unit control section 480 further includes an interface section 488 that is interposed between various hardware (i.e., the specimen suction section 450, the device mechanism section 430, the sample preparation section 440, and the FCM detection section 410) and the bus 485.

[0032] The A / D conversion unit 482 samples the analog signal output from the analog processing unit 420 at a predetermined sampling rate (for example, sampling at 1024 points at 10 nanosecond intervals, sampling at 128 points at 80 nanosecond intervals, or sampling at 64 points at 160 nanosecond intervals) and converts it into a digital signal. The A / D conversion unit 482 converts the analog signal from the start of sample measurement to the end of measurement into a digital signal. When multiple types of analog signals (for example, analog signals corresponding to forward scattered light intensity, side scattered light intensity, and fluorescence intensity) are generated by measurement in a certain measurement channel, the A / D conversion unit 482 converts each analog signal from the start of measurement to the end of measurement into a digital signal. As described with reference to FIG. 4, three types of analog signals (i.e., forward scattered light signal, side scattered light signal, and fluorescence signal) are input to the A / D conversion unit 482 via multiple corresponding signal transmission paths 421. The A / D conversion unit 482 converts each of the analog signals input from the multiple signal transmission paths 421 into a digital signal.

[0033] FIG. 5 is a schematic diagram for explaining the processing of analog signal sampling by the A / D conversion unit 482. When a measurement sample containing cells C is flowed through the flow cell 4113 and light is irradiated onto the flow cell 4113, forward scattered light is generated in the forward direction relative to the direction of light propagation. Similarly, side scattered light and side fluorescent light are generated to the sides relative to the direction of light propagation. The forward scattered light is received by the light receiving element 4116, and a signal corresponding to the amount of received light is output. The side scattered light is received by the light receiving element 4121, and a signal corresponding to the amount of received light is output. The side fluorescent light is received by the light receiving element 4122, and a signal corresponding to the amount of received light is output. As multiple cells contained in the measurement sample pass through the flow cell 4113, analog signals representing changes in the signal over time are output from the light receiving elements 4116, 4121, and 4122. The analog signal corresponding to forward scattered light is called the "forward scattered light signal," the analog signal corresponding to side scattered light is called the "side scattered light signal," and the analog signal corresponding to side fluorescence is called the "fluorescence signal." One pulse of each analog signal corresponds to one cell.

[0034] The analog signals are input to the A / D converter 482. The A / D converter 482 samples the forward scattered light signal, the side scattered light signal, and the fluorescent light signal, starting from the point when the level of the forward scattered light signal reaches a level set as a predetermined threshold, among the analog signals input from the light receiving elements 4116, 4121, and 4122. The A / D converter 482 samples each analog signal at a predetermined sampling rate (for example, sampling at 1024 points at 10 nanosecond intervals, sampling at 128 points at 80 nanosecond intervals, or sampling at 64 points at 160 nanosecond intervals). The sampling time is fixed regardless of the pulse size. The sampling time is set to be longer than the time it takes for the analog signal level to rise and fall when one cell passes through the beam spot of the flow cell 4113. As a result, matrix data is obtained as a digital signal corresponding to one cell, with elements each representing the analog signal level at multiple points in time. In this way, the A / D conversion unit 482 generates a digital signal of forward scattered light, a digital signal of side scattered light, and a digital signal of side fluorescent light corresponding to one cell. The A / D conversion is repeated until the number of cells from which digital signals have been obtained reaches a predetermined number, or until a predetermined time has passed since the measurement sample began to flow through the flow cell 4113. As a result, as shown in FIG. 5, digital signals are obtained for N cells in the measurement sample by digitizing the waveforms of the analog signals of each cell. In this specification, the collection of sampling data for each cell contained in the digital signal (in the example of FIG. 5, a collection of 1024 digital values ​​from t=0 ns to t=10240 ns) is referred to as waveform data.

[0035] An index for identifying each cell is assigned to each piece of waveform data generated by the A / D conversion unit 482. For example, integers from 1 to N are assigned to the index in the order in which the waveform data is generated, and the same index is assigned to waveform data of forward scattered light, waveform data of side scattered light, and waveform data of side fluorescent light obtained from the same cell. The index corresponds to the cell ID included in the cell data structure described below.

[0036] Since one waveform data corresponds to one cell, the index corresponds to the measured cell. By assigning the same index to waveform data corresponding to the same cell, the deep learning algorithm described below can analyze the waveform data of forward scattered light, waveform data of side scattered light, and waveform data of fluorescence corresponding to each cell and classify the cell type.

[0037] In addition to generating waveform data corresponding to each cell, the A / D conversion unit 482 calculates the peak value of the pulse of each signal to generate feature parameter data. FIG. 6 is a schematic diagram showing an example of feature parameter data. In parallel with the process of converting the analog signal of each cell into a digital value to generate waveform data, the A / D conversion unit 482 stores the maximum digital value contained in the waveform data of each cell, starting from the first column of the feature parameter data. Column N stores the value of the cell for which waveform data was generated Nth. The column position corresponds to the index assigned to the beginning of the waveform data. In other words, the digital value corresponding to the waveform data with index "N" is stored in column N. The maximum digital value corresponds to the peak height of the pulse of the analog signal of the cell. Therefore, by extracting the maximum digital value contained in the waveform data of each cell for the forward scattered light signal, side scattered light signal, and side fluorescent light signal, the peak value of the pulse of the forward scattered light signal (referred to as FSCP), the peak value of the pulse of the side scattered light signal (referred to as SSCP), and the peak value of the pulse of the side fluorescent light signal (referred to as SFLP) can be obtained for each cell.

[0038] 2, the A / D conversion unit 482 inputs the generated digital signal and feature parameter data to the bus 485. The bus controller 4850 transmits the digital signal output from the A / D conversion unit 482 to the RAM 4834 by, for example, DMA (Direct Memory Access) transfer. The RAM 4834 stores the digital signal and feature parameter data.

[0039] The processor 4831 is connected to the interface unit 489, the interface unit 488, the interface unit 484, the RAM 4834, and the storage unit 4835 via the bus 485. The processor 4831 is connected to the parallel processing processor 4833 via the bus 485. The processing unit 300 is connected to each component of the measuring unit 400 via the interface unit 489 and the bus 485. The bus 485 is, for example, a transmission path having a data transfer rate of several hundred MB / s or more. The bus 485 may be, for example, a transmission path having a data transfer rate of 1 GB / s or more. The bus 485 transfers data based on, for example, the PCI-Express or PCI-X standard.

[0040] The processor 4831 uses the parallel processing processor 4833 to execute analysis software 4835 a stored in the memory unit 4835 to analyze the digital signal.

[0041] The processor 4831 is, for example, a central processing unit (CPU). The parallel processing processor 4833 executes multiple arithmetic processes in parallel, which are at least a part of the processing related to the analysis of waveform data. The parallel processing processor 4833 is, for example, a graphics processing unit (GPU), a field programmable gate array (FPGA), or an application specific integrated circuit (ASIC). When the parallel processing processor 4833 is an FPGA, the parallel processing processor 4833 may be pre-programmed with arithmetic processes related to the deep learning algorithm 60, which are implemented by the analysis software 4835a. When the parallel processing processor 4833 is an ASIC, the parallel processing processor 4833 may be pre-programmed with a circuit for executing arithmetic processes related to the deep learning algorithm 60, or may have a programmable module built in addition to such a built-in circuit. For example, NVIDIA's GeForce, Quadro, TITAN, or Jetson is preferably used as the parallel processing processor 4833.

[0042] FIG. 7 is a diagram illustrating the calculation process of the deep learning algorithm 60 of the analysis software 4835a.

[0043] The deep learning algorithm 60 is configured by a neural network including multiple hidden layers. The neural network is preferably a convolutional neural network (CNN) having a convolutional layer. The number of nodes in the input layer 60a of the neural network corresponds to the number of sequences included in the input waveform data of one cell. In the example of FIG. 6, the number of nodes in the input layer 60a corresponds to the number of 1024 sequences x 3 of the waveform data 86a, 86b, and 86c of one cell.

[0044] The output layer 60b of the neural network has nodes in a number corresponding to the cell types to be analyzed. In the example of Figure 8, there are nine cell types to be analyzed: "neutrophils (NEUT)," "lymphocytes (LYMPH)," "monocytes (MONO)," "eosinophils (EO)," "basophils (BASO)," "immature granulocytes (IG)," "blasts (Blast)," "abnormal lymphocytes (Abn LYMPH)," and "none." In this case, the number of nodes in the output layer 60b is nine.

[0045] When waveform data is input to input layer 60a of the neural network that constitutes deep learning algorithm 60, output layer 60b outputs the probability that the cell corresponding to the waveform data corresponds to each cell type as classification information 82 that classifies cells into multiple cell types. The data output from output layer 60b includes the cell ID, cell type identification information (label value in FIG. 8), and the numerical value of the probability corresponding to each cell type.

[0046] The processor 4831 stores the cell classification information 82 obtained by the deep learning algorithm 60 in the RAM 4834. The classification information 82 and feature parameter data of each cell are transmitted to the processing unit 300 via the interface section 489.

[0047] 3. Configuration of Processing Unit 300 The configuration of the processing unit 300 will be described with reference to Fig. 9. The processing unit 300 is connected to the processor 4831 of the measurement unit 400 via an interface unit 3006 and a bus 3003, and can receive the classification information 82 and feature parameter data generated by the measurement unit 400. The interface unit 3006 is, for example, a USB interface.

[0048] The processing unit 300 includes a processor 3001, a bus 3003, a memory 3004, an interface 3006, a display 3015, and an operation unit 3016. The processing unit 300 is configured from a general personal computer as hardware, and functions as the processing unit of the cell analyzer 100 by executing a dedicated program stored in the memory 3004.

[0049] The processor 3001 is a CPU and is capable of executing programs stored in the storage unit 3004 .

[0050] The storage unit 3004 includes a hard disk drive. The storage unit 3004 stores at least a program for processing the cell classification information 82 transmitted from the measurement unit 400 and generating the test results of the specimen.

[0051] The display unit 3015 includes a computer screen. The display unit 3015 is connected to the processor 3001 via the interface unit 3006 and the bus 3003. The display unit 3015 can receive image signals input from the processor 3001 and display the inspection results.

[0052] The operation unit 3016 is equipped with a pointing device including a keyboard, a mouse, or a touch panel. By operating the operation unit 3016, users such as doctors and laboratory technicians can input measurement orders to the cell analyzer 100 and input measurement instructions according to the measurement orders. The operation unit 3016 can also accept instructions from the user to display test results. By operating the operation unit 3016, the user can view various information related to the test results, such as graphs, charts, and flag information assigned to samples.

[0053] 4. Operation of Cell Analysis Device 100 The operation of analyzing a sample by the cell analyzer 100 will be described with reference to FIG.

[0054] When the processor 3001 of the processing unit 300 receives a measurement order and a measurement instruction from a user via the operation unit 3016, the processor 3001 transmits a measurement command to the measurement unit 400 (step S1).

[0055] Upon receiving the measurement command, the processor 4831 of the measurement unit 400 starts measuring the sample. The processor 4831 causes the sample aspirator 450 to aspirate the sample from the blood collection tube T (step S10). Next, the processor 4831 causes the sample aspirator 450 to dispense the aspirated sample into one of the reaction chambers 440a to 440e of the sample preparation section 440. The measurement command sent from the processing unit 300 in step S1 includes information about the measurement channel for which measurement is requested by the measurement order. The processor 4831 controls the sample aspirator 450 to discharge the sample into the reaction chamber of the corresponding measurement channel based on the measurement channel information included in the measurement command.

[0056] Processor 4831 causes sample preparation unit 440 to prepare a measurement sample (step S11). In step S11, sample preparation unit 440 receives a command from processor 4831 and supplies reagents (hemolytic agent and staining solution) to the reaction chamber into which the specimen has been discharged, and mixes the specimen with the reagents. As a result, a measurement sample is prepared in which red blood cells are hemolyzed by the hemolytic agent in the reaction chamber and cells targeted by the measurement channel, such as white blood cells and reticulocytes, are stained by the staining agent.

[0057] Processor 4831 causes FCM detection unit 410 to measure the prepared measurement sample (step S12). In step S12, processor 4831 controls device mechanism unit 430 to send the measurement sample in the reaction chamber of sample preparation unit 440 to FCM detection unit 410. The reaction chamber and FCM detection unit 410 are connected by a flow path, and the measurement sample sent from the reaction chamber flows through flow cell 4113 and is irradiated with laser light by light source 4111 (see FIG. 4).

[0058] When a measurement sample is supplied to the flow cell 4113, a forward scattered light signal, a side scattered light signal, and a side fluorescent light signal are input to the A / D conversion unit 482. The A / D conversion unit 482 generates a digital signal that is a collection of waveform data corresponding to each cell, and feature parameter data that stores the maximum value contained in the waveform data of each cell. The method for generating the waveform data and feature parameters has already been described.

[0059] The processor 4831 controls the bus controller 4850 to cause the waveform data and feature parameter data generated by the A / D conversion unit 482 to be loaded into the RAM 4834 by DMA transfer. By DMA transfer, the waveform data is transferred directly to the RAM 4834 without going through the processor 4831. The waveform data is stored in the RAM 4834.

[0060] The processor 4831 uses the deep learning algorithm 60 to perform cell classification based on the generated waveform data and generate classification information (step S13).

[0061] The processor 4831 transmits the classification information 82 and the characteristic parameter data of each cell obtained as a result of step S14 to the processing unit 300 (step S14).

[0062] When the processor 3001 of the processing unit 300 receives the classification information 82 and the feature parameter data from the measurement unit (step S2), it analyzes the classification information 82 using a program stored in the storage unit 3004 and generates test result data for the specimen (step S3). The test result data is stored in the storage unit 3004. The processing of step S3 will be described later.

[0063] Processor 3001 displays the test results on display unit 3015 (step S4). The processing of step S4 will be described later.

[0064] The processor 3001 transmits the test results to the host computer 500 (step S5). The process of step S5 will be described later.

[0065] The host computer 500 receives the test results sent from the processor 3001 of the processing unit 300 (step S21), which completes the series of processes.

[0066] [5. Test result data generation process] FIG. 11 is a flowchart showing the details of step S3 (test result data generation process) executed by the processor 3001 of the processing unit 300.

[0067] The processor 3001 determines the main cell type of the cells based on the classification information 82 received from the measurement unit 400 (step S31). Specifically, the processor 3001 determines the cell type with the highest probability as the main cell type based on the probability of each cell type included in the classification information 82. For example, if the probability data obtained for a certain cell is as follows, the cell type with the highest probability is a neutrophil, and therefore the main cell type of that cell is a neutrophil.

[0068] [Table 1]

[0069] Next, processor 4831 counts the number of cells contained in the specimen for each cell type based on the determined main cell type (step S32). In the processing of step S32, the number of cells is counted for each cell type based on the cell type information of each cell. For example, when WDF is set as the measurement channel, there are nine cell types to be classified, as shown in FIG. 8. For example, if there are M cells whose main cell type is neutrophils, processor 3001 generates a count result in which the number of neutrophils is M. Processor 3001 performs similar processing on lymphocytes, monocytes, eosinophils, basophils, immature granulocytes, blasts, and abnormal lymphocytes, and generates a count result for each cell type.

[0070] The processor 3001 may further determine whether a sample is normal or abnormal based on the count value of each cell type. For example, because blast cells are not found in the peripheral blood of healthy individuals, if a sample contains a predetermined number or more of cells determined to be blast cells as the main cell type, the sample is suspected of having some kind of abnormality. Furthermore, because immature granulocytes are not normally found in the blood of healthy individuals, if a sample contains a predetermined number or more of cells determined to be immature granulocytes as the main cell type, the sample is also suspected of having some kind of abnormality. The processor 3001 may compare the count value (absolute value) of each cell type with a predetermined numerical range or threshold, and if the count value falls outside the numerical range or exceeds the threshold, add a flag to the analysis result indicating a suspected abnormality.

[0071] The processor 3001 may further determine the cell content ratio based on the count value of each cell type. For example, when WDF is set as the measurement channel, the processor 3001 determines the ratio of five white blood cell subclasses: neutrophils, lymphocytes, monocytes, eosinophils, and basophils. These five white blood cell subclasses are contained in predetermined proportions in the blood of healthy individuals. The processor 3001 may determine whether the sample is normal or abnormal by comparing the ratio of the white blood cell subclasses with a predetermined numerical range or threshold. The processor 3001 may add a flag to the analysis result indicating that the sample is abnormal if a specific white blood cell ratio falls outside a predetermined numerical range or exceeds a threshold.

[0072] The processor 3001 generates the above-mentioned test result data in association with the cell ID (step S33). Fig. 12 is a diagram schematically showing a first example of the data structure of the test result data generated by the processor 3001.

[0073] The test result data is configured in a relational database and includes multiple data items (columns). In the example shown in Fig. 12, the data items include a subject ID, a specimen ID, a measurement result, an identification code of the measurement channel, a cell ID, identification information of the cell type, and a probability for each cell type.

[0074] The subject ID data stores the subject ID, which is information for identifying the subject from whom the sample was collected. The subject ID is, for example, a multi-digit character string consisting of a combination of numbers and letters.

[0075] The sample ID data stores a sample ID, which is information for identifying a sample. The sample ID is, for example, a multi-digit character string consisting of a combination of numbers and letters.

[0076] The measurement result data stores the measurement results obtained by the processor 3001 analyzing the measurement data based on the main cell type in step S32. As described above, the measurement result includes the cell count result based on the cell type (main cell type) with the highest probability. If the measurement channel is WDF, the measurement result includes the number and percentage of each white blood cell subclass (monocytes, neutrophils, lymphocytes, eosinophils, and basophils). If the measurement channel is RET, the measurement result includes the reticulocyte count. If the measurement channel is WPC, the measurement result includes the hematopoietic progenitor cell count. If the measurement channel is WNR, the measurement result includes the white blood cell count and the nucleated red blood cell count.

[0077] The measurement channel data stores information indicating the measurement channel used for the measurement of the sample. In the example of the measurement unit 400 of this embodiment, the measurement channels are DIFF, RET, WPC, PLT-F, and WNR. The measurement channel data item stores character strings such as "DIFF," "RET," "WPC," "PLT-F," and "WNR" corresponding to the measurement channel used. The information indicating the measurement channel may be a character string indicating the measurement channel, or may be numbers or letters assigned to each measurement channel. For example, numbers may be assigned, such as DIFF=1, RET=2, WPC=3, PLT-F=4, and WNR=5, and the number corresponding to the measurement channel used may be stored. The measurement channel data stores one or more values ​​for one sample depending on the number of measurement channels used for the measurement.

[0078] Multiple cell ID data are created according to the number of cells detected by measurement on one measurement channel. Cell IDs are natural numbers such as 1, 2, 3, ..., N. The cell ID matches the index added to the beginning of the waveform data corresponding to each cell mentioned above (see Figure 5) and the column order of the feature parameter data (see Figure 6).

[0079] The feature parameter data stores the values ​​of the forward scattered light peak (FSCP), side scattered light peak (SSCP), and side fluorescent light peak (SFLP) of the corresponding cell. In the example of FIG. 12, three sets of feature parameter data are created for one cell ID: FSCP, SSCP, and SFLP. As described with reference to FIG. 6, the column order of the feature parameter data corresponds to the cell ID. For example, the first column (first column) of the feature parameter data stores feature parameters corresponding to a cell with cell ID "1." Therefore, the FSCP data corresponding to cell ID=1 stores the value "59" of the first column of the FSCP matrix data among the feature parameter data. Similarly, the SSCP and SFLP store the value "30" of the first column of the SSCP matrix data and the value "134" of the first column of the SFLP matrix data, respectively.

[0080] The cell type data stores information indicating the cell type to be analyzed. The information indicating the cell type is, for example, a label value that can identify the cell type, as shown in Figure 8. Multiple pieces of cell type data are created depending on the number of cell types to be analyzed in the measurement channel. For example, if there are nine types of cell types to be analyzed in the measurement channel, including "none," as shown in Figure 8, nine pieces of cell type data are created. Note that the identification information may be a character string such as "neutrophil" or "NEUT" instead of a label value.

[0081] One piece of probability data is created for each cell type data. The probability data stores the probability value output by the deep learning algorithm 60 for the corresponding cell ID and identification information in the form of numerical data. For example, if the classification information 82 of the target cell is as shown in Table 1 above, "90%" is stored in the variable corresponding to neutrophils, "10%" is stored in the variable corresponding to lymphocytes, and "0%" is stored in the variables corresponding to other identification information.

[0082] The processor 3001 generates the test result data shown in FIG.

[0083] Fig. 13 is a diagram schematically showing a second example of test result data. Compared to Fig. 12, the data structure of the test result data shown in Fig. 13 has added data that stores a main flag for each piece of identification information. In the case of the data structure of Fig. 13, the classification information is made up of identification information, a probability, and a main flag.

[0084] The main flag data stores a value of 0 or 1. The processor 3001 stores 1 in the main flag corresponding to the cell type with the highest probability (main cell type) and stores 0 in the main flags corresponding to the other cell types. This configuration eliminates the need for the processor 3001 to identify the main cell type for each cell based on the probability each time the main cell type information is used in a calculation or output, thereby speeding up processing. For example, in a scattergram (described later), the plots of each cell are displayed in different colors depending on the cell type. In such a case, the processor 3001 can determine the color of the plot of each cell by referencing the main flag, eliminating the need to identify the main cell type based on the probability of each cell. In the second example, when the deep learning algorithm 60 generates the classification information 82, it may execute a process of setting a numerical value in the main flag along with the classification information 82.

[0085] Figure 14 is a diagram schematically showing a third example of test result data. Compared to Figure 12, the test result data shown in Figure 14 has one main cell type data added for each cell ID. In the case of the data structure of Figure 14, the classification information is composed of identification information, probability, and identification information indicating the main cell type.

[0086] The main cell type data stores the identification information of the cell type with the highest probability (main cell type). In the second example of Figure 13, it is necessary to search the column storing the main flag to identify the main cell type of each cell, but in the third example of Figure 14, the main cell type can be identified by referencing the main cell type data, which enables further speedup of processing.

[0087] Fig. 15 is a diagram showing a fourth example of test result data. Compared to Fig. 12, the test result data shown in Fig. 15 has research flag data added to it corresponding to each piece of identification information. In the case of the data structure of Fig. 15, the classification information is made up of identification information, probability, and research flag.

[0088] The research flag data stores a value of 0 or 1. The research flag is a flag used to distinguish between cell types displayed as test results for reportable items and cell types displayed as auxiliary research items on the result display screen described below. The processor 3001 stores "1" in the research flag corresponding to a cell type whose probability is below a predetermined threshold, and stores "0" in the research flag for a cell type whose probability is greater than this threshold. When the research flag is set to 1, when test result data is displayed, information indicating that the probability, counting results, etc. for this cell type are preferably used for research purposes can be added.

[0089] Fig. 16 is a diagram schematically showing a fifth example of test result data. In the test result data shown in Fig. 16, ranking data is added to each piece of identification information, as compared to Fig. 12. In the case of the data structure of Fig. 16, the classification information is composed of identification information, probability, and ranking.

[0090] The ranking data stores numbers from 1 to m (m: the number of cell types to be analyzed) assigned to all cell types associated with the cell ID in descending order of probability. Processor 4831 calculates the ranking in descending order of probability and stores the calculated ranking in the ranking data for each cell type. In this case, by referring to the ranking data, processor 3001 can smoothly grasp not only the cell type with the highest probability, but also, for example, the cell type with the second highest probability.

[0091] [6. Test result display process] FIG. 17 is a flowchart illustrating a first example of the test result display process (step S4) executed by the processor 3001.

[0092] The processor 3001 displays the sample list screen 700 (sample explorer screen) (step S410). The processor 3001 displays the details screen 800 (browser screen) in response to a user operation (step S411). The processor 3001 displays the analysis screen 900 including a scattergram in response to a display instruction from the user (step S412). The processor 3001 reads out classification information of the cell corresponding to the dot selected in the scattergram 901 displayed on the analysis screen 900, and displays the read out classification information (step S413).

[0093] 18 shows an example of a sample list screen 700. The sample list screen 700 includes a toolbar 710 and a data display area 720. The toolbar 710 displays multiple icons for performing specific operations on the screen. Specifically, the toolbar 710 includes a sample list screen icon 710a for calling the sample list screen 700 (SE), a details icon 710b for calling the details screen 800, a measurement registration icon 710c for calling the measurement order input screen for entering a measurement order, and a validate icon 710d for validating test results.

[0094] A data display area 720 in which test result data is displayed is provided below the toolbar 710. The toolbar 710 is always displayed at the top of the screen regardless of the content displayed in the data display area 720. A sample list display area 931 and a measurement result display area 932 are displayed in the data display area 720 of the sample list screen 700.

[0095] The sample list display area 931 has items such as subject ID, sample ID, date and time, etc. The sample list display area 931 displays test result data identified by the subject ID and sample ID in list form.

[0096] The measurement result display area 932 displays the measurement results based on the sample ID selected in the sample list display area 931. The measurement results displayed in the measurement result display area 932 are counting results and ratio information based on the main cell type.

[0097] When the user selects the details icon 710b on the sample list screen 700 or selects (for example, double-clicks) the record of one sample in the sample list display area 931, the details screen 800 is displayed.

[0098] FIG. 19 is a diagram showing an example of a details screen 800. The details screen 800 includes a toolbar 810 (710) and a data display area 820. The toolbar 810 is the same as that displayed on the sample list screen 700. The data display area 820 displays a measurement result area 820a for displaying measurement results, a flag area 820b for displaying flag information, and a graph area 820c for displaying graphs. The measurement result area 820a displays the measurement results of all measurement channels in which measurements were performed on the sample to be displayed. The flag area 820b displays a flag (suspect message) if a flag has been set for the sample based on the measurement results. The graph area 820c displays a graph corresponding to each measurement channel.

[0099] 19, the graph area 820c displays a WDF scattergram, a WNR scattergram, a WPC scattergram, a RET scattergram, a PLT-F scattergram, an RBC histogram, and a PLT histogram. The scattergrams displayed in the graph area 820c are created and displayed based on the above-mentioned characteristic parameters.

[0100] When the user wants to validate the test result data by referring to the analysis results displayed on the details screen 800, the user operates the validate icon 710d. This validates the test result data. When the user does not want to validate the test result data, the user can operate the measurement registration icon 710c, for example, to input a retest order.

[0101] To check the classification information of the cells, the user can display the analysis screen 900. When the user operates (e.g., double-clicks) any graph in the graph area 820c of the details screen 800, the analysis screen 900 pops up and is overlaid on the details screen 800.

[0102] Fig. 20 is an example of the analysis screen 900. Fig. 20 illustrates an example of the analysis screen 900 that is displayed when the user selects a WDF scattergram.

[0103] The analysis screen 900 displays a scattergram 901 and a detailed display area 902. The scattergram 901 shown in FIG. 20 is displayed based on the characteristic parameters of cells associated with the measurement channels included in the test result data. In the scattergram 901 shown in FIG. 20, the horizontal axis represents the side scattered light signal peak value (SSCP) and the vertical axis represents the side fluorescent light signal peak value (SFLP). For each cell identified by a cell ID, the processor 3001 determines coordinates in the scattergram 901 based on the SSCP and SFLP, and plots the cell by drawing dots at the determined coordinates. The processor 3001 performs the same processing on the data of all cells under the measurement channel corresponding to the displayed scattergram 901, and plots the cells on the scattergram 901.

[0104] The processor 3001 sets the color of the dots based on the classification information of the plotted cells. Specifically, the processor 3001 displays cells of different cell types with dots of different colors. For example, in the scattergram 901 corresponding to the WDF channel, neutrophils are displayed as a first color (e.g., blue), lymphocytes as a second color (e.g., purple), monocytes as a third color (e.g., green), eosinophils as a fourth color (e.g., red), and basophils as a fifth color (e.g., yellow).

[0105] In conventional white blood cell classification in hemocytometers, cells are plotted on a coordinate plane, such as the scattergram 901 shown in FIG. 20, and clustering analysis is used to classify the cells into multiple populations based on the coordinates of each cell. Therefore, cells plotted at the same coordinates are classified as cells of the same type. In contrast, in this embodiment, unlike conventional white blood cell classification, cell characteristic parameters (SSCP and SFLP) are only used to plot the cells on the scattergram 901. As described above, cell classification in this embodiment involves analyzing the waveform data of individual cells using the deep learning algorithm 60, thereby identifying individual cell types without relying on clustering techniques. In other words, the coordinates of the cell plots in the scattergram 901 are fragmentary information representing the distribution of cells contained in the measurement sample based on parameters representing the cell characteristics, and the coordinates are not used to identify the cell type. Therefore, even if multiple cells are plotted at the same coordinates, these cells may be classified as different cells by the deep learning algorithm 60. When different types of cells are plotted at the same coordinates, the processor 3001 draws dots of a color corresponding to the most prevalent main cell type based on the main cell types of the multiple cells plotted at the same coordinates.

[0106] In step S413 of the flowchart in FIG. 17, when the user performs an operation to select a dot on the scattergram 901 (e.g., by double-clicking), the processor 3001 pops up a detailed display area 902. The detailed display area 902 includes a classification information display area 902a, which displays classification information of the cells included in the dot selected in the scattergram 901. FIG. 20 shows an example in which multiple cell types and the probabilities corresponding to each cell type are displayed. The classification information display area 902a is provided with a scroll bar for changing the display range within the classification information display area 902a so that all cell types and their probabilities can be displayed. When multiple cells are plotted at the selected dot on the scattergram 901, a classification information display area 902a for each of the multiple cells is displayed in the detailed display area 902. The user can scroll through the classification information by operating the scroll bar.

[0107] 20, classification information for the cells contained in the selected dot is displayed in a pop-up, but a classification information display area 902a may also be provided on the analysis screen 900, as shown in Fig. 21. In this case, the classification information display area 902a is blank until a dot is selected, and once a dot is selected, classification information for the cells corresponding to the selected dot is displayed in the classification information display area 902a.

[0108] In the above example, the detailed display area 902 is displayed as a pop-up in response to an operation (e.g., double-clicking) on ​​a dot on the scattergram 901, but the operation for calling up the detailed display area 902 can include various operations as long as it is possible to specify any coordinate on the scattergram 901. For example, the detailed display area 902 may be displayed simply by placing the cursor on the scattergram 901, in which case placing the cursor on the scattergram 901 is the operation for calling up the detailed display area 902. Furthermore, instead of double-clicking, the operation may be a long press, a right click, or pressing a predetermined key on the keyboard or software keys (e.g., the Enter key) with the cursor placed on the coordinate.

[0109] 20 and 21, the user can refer to the classification information display area 902a to check the classification information of the cell corresponding to the dot selected on the scattergram 901. Furthermore, by changing the dot selected on the scattergram 901, the user can dynamically switch the display content of the classification information display area 902a and check the display content continuously.

[0110] In addition, the classification information display area 902a may display other information as classification information, not limited to the probability of each cell type. For example, if the test result data has the configuration of FIG. 13, the value of the main flag for each cell type may be displayed. Furthermore, if the test result data has the configuration of FIG. 14, a mark may be displayed on the main cell type (the cell type with the highest probability), or the main cell type may be displayed separately. Furthermore, if the test result data has the configuration of FIG. 15, the value of the research flag for each cell type may be displayed. Furthermore, if the test result data has the configuration of FIG. 16, the ranking of each cell type may be displayed.

[0111] In Figures 20 and 21, the probability of each cell type is displayed numerically. Alternatively, a graph corresponding to the probability of each cell type may be displayed, as shown in Figures 22 and 23. Figure 22 is a diagram showing another example of the analysis screen 900. In the example of Figure 22, the classification information display area 902a displays cell probability information for each cell corresponding to the selected dot as a 100% stacked bar chart. This 100% stacked bar chart is displayed corresponding to each cell ID, and the probability values ​​for each identification information are displayed in a stacked format to total 100%. This allows the user to visually grasp the probability of each cell type.

[0112] Fig. 23 is a diagram showing another example of the analysis screen 900. On the analysis screen 900 of Fig. 23, a scattergram 901, a detailed display area 902, an enlargement button 903, and an enlarged view 904 are displayed.

[0113] When the user operates the enlargement button 903, a gate 901a of the reference range is displayed on the scattergram 901. The gate 901a is used to select a desired dot on the scattergram 901 by specifying a range. The user can change the position of the gate 901a on the scattergram 901 by performing an operation such as dragging. The enlarged view 904 displays an enlarged view of the portion of the scattergram 901 within the gate 901a. When the user performs an operation to select a dot on the enlarged view 904, the classification information of the selected cell is displayed in a classification information display area 902a in the detailed display area 902, as in Figures 20 and 21.

[0114] 23, the user can set a gate 901a on the scattergram 901 and display an enlarged view 904, thereby easily selecting a target cell in the enlarged view 904. This allows the user to smoothly check the classification information even for cells in an area where cell dots are densely packed in the scattergram 901.

[0115] The size of gate 901a may be changed by performing an operation such as dragging on the boundary line of gate 901a. Also, the size of gate 901a may be changed in stages each time enlargement button 903 is operated.

[0116] Although the examples in FIGS. 20 to 23 show an example in which one dot on the scattergram 901 is selected, the number of dots that can be selected at one time is not limited to one. For example, multiple dots may be selected by clicking and holding down a mouse on any point on the scattergram 901 and dragging the cursor to specify a range on the scattergram 901. However, if multiple dots are selected at once, the number of cells to be displayed may be too large, and displaying the classification information for each cell individually may be difficult for the user to view. Therefore, as will be described later with reference to FIG. 29, the display method may be switched depending on the number of selected dots. For example, if the number of selected dots is equal to or less than a predetermined threshold, the classification information for the cells may be displayed individually, whereas if the number of selected dots is greater than the predetermined threshold, the classification information for the cells may be displayed statistically instead of individually.

[0117] Figure 24 is a flowchart showing a second example of the test result display process. In Figure 24, step S413 has been changed compared to the flowchart in Figure 17. In step S413 in Figure 24, classification information for the cell corresponding to the dot selected in the scattergram 901 is read out and displayed on the analysis screen 900 as statistical information.

[0118] In the first example of the test result display process in Fig. 17, as shown in the examples of Figs. 20 to 23, the classification information of the cells corresponding to the selected dots is displayed individually for each cell. Alternatively, in the examples of Figs. 25 to 28 shown below, the classification information of multiple cells corresponding to the selected dots is tallied and statistical information is displayed. Examples of the analysis screen 900 are shown in Figs. 25 to 28.

[0119] An analysis screen 900 in FIG. 25 displays a scattergram 901, an enlargement button 903, an enlarged view 904, and a detailed display area 905.

[0120] The detailed display area 905 displays statistical information on the probability and number of each cell type using a histogram 905a based on the classification information of all cells within the gate 901a. The histogram 905a is frequency distribution information obtained by aggregating all cells within the gate 901a by the probability of that cell type. In the histogram 905a, the horizontal axis represents the probability and the vertical axis represents the number of cells. The number shown in the histogram 905a also includes the count results of cells that are not the main cell type (cells with a low probability of being that cell type). The detailed display area 905 is provided with a scroll bar that allows the display range within the detailed display area 905 to be changed so that the histograms 905a of all cell types can be displayed.

[0121] 25, the user can grasp the probability distribution of each cell type for all cells in the gate 901a by displaying histograms 905a for each cell type in the detailed display area 905. For example, in conventional clustering methods using scattergrams, the distribution ranges of multiple cell types may overlap on the scattergram 901. In contrast, according to this embodiment, the frequency distribution of cell types can be confirmed by setting gate 901a in the overlapping range of such distributions and referring to histogram 905a. This allows the user to make detailed judgments about the specimen, for example, to determine the likelihood of cells of each cell type being included in the overlapping range of distributions.

[0122] 26 is a diagram showing another example of the analysis screen 900. The analysis screen 900 displays a scattergram 901, an enlargement button 903, an enlarged view 904, a cell type selection area 906, and a three-dimensional histogram 907 as statistical information.

[0123] The cell type selection area 906 is provided with check boxes for selecting cell types. The cell type selection area 906 is provided with a scroll bar for changing the display range within the cell type selection area 906 so that all cell types can be displayed. The three-dimensional histogram 907 is frequency distribution information obtained by aggregating all cells within the gate 901a by probability of the selected cell type. The three-dimensional histogram 907 displays histograms for each selected cell type in a three-dimensional manner, arranged front to back. In the example of FIG. 26, lymphocytes and monocytes are selected in the cell type selection area 906, and therefore the three-dimensional histogram 907 displays the histograms for lymphocytes and monocytes together.

[0124] According to the analysis screen 900 of Fig. 26, the user can simultaneously check the frequency distributions of two or more cell types by selecting two or more cell types and referring to the three-dimensional histogram 907. In the example of Fig. 25, two histograms 905a were displayed simultaneously on the screen, but in the example of Fig. 26, by selecting three or more cell types, three or more histograms are displayed on the screen. Thus, according to the screen of Fig. 26, the user can simultaneously grasp even more frequency distributions than with the screen of Fig. 25. Furthermore, the user can easily compare frequencies for each probability.

[0125] The three-dimensional histogram 907 in FIG. 26 is a histogram in which multiple histograms 905a shown in FIG. 25 are arranged in the depth direction, but instead, the distribution of the probabilities of multiple cell types may be displayed using a surface plot or the like.

[0126] Fig. 27 is a diagram showing another example of the analysis screen 900. On the analysis screen 900 of Fig. 27, a scattergram 901, an enlargement button 903, an enlarged view 904, and a number display area 908 are displayed.

[0127] In the number display area 908, the count values ​​of all cells within the gate 901a, totaling the cells with a probability of more than 0% for each cell type, are displayed as statistical information in a bar graph. In this case, because multiple cells within the gate 901a are counted overlapping, the total count values ​​in the number display area 908 will be greater than the actual number of cells within the gate 901a. For example, suppose that the gate 901a in FIG. 27 contains one cell with probability information of 90% monocyte, 5% neutrophil, and 5% immature granulocyte. In this case, the cell is counted as one count for each of the monocyte, neutrophil, and immature granulocyte. In other words, if the probability of belonging to multiple cell types is 0% or greater, the number of that cell is counted overlapping with the number of multiple cell types. Note that the counting method is not limited to this, and the number may be counted by expressing it as a decimal of 1 or less according to the probability. For example, as in the example above, if the reference range of gate 901a contains one cell with probability information of 90% monocyte, 5% neutrophil, and 5% immature granulocyte, the count may be calculated assuming that the range contains 0.9 monocytes, 0.05 neutrophils, and 0.05 immature granulocytes.

[0128] If the count result of a cell type is less than a predetermined threshold (for example, 10), the count value of the cell type is added to the bar graph in the number display area 908. In the example of Fig. 27, the count value of immature granulocytes is 5, so "5" is added to the bar graph of immature granulocytes. In addition, the number display area 908 is provided with a scroll bar for changing the display range in the number display area 908 so that all cell types can be displayed.

[0129] According to the analysis screen 900 of FIG. 27, the user can check the number of each cell type in the gate 901a by referring to the number display area 908. Cells with a probability exceeding a predetermined percentage are counted as one, which is convenient for checking the majority and minority cell types. Furthermore, for cell types with small count values ​​in the number display area 908, the count value is displayed as an auxiliary display to confirm the degree of the count value, so the count value of the cell type with the small count value can be clearly confirmed. This allows the user to smoothly determine the possibility that a rare cell type may be included in the sample.

[0130] Although the count value of the cell types with a probability greater than 0% is displayed in the number display area 908, the threshold value for counting is not limited to 0% and may be set to any value greater than 0% and less than 100%. Also, a user interface that can accept the threshold value for counting (for example, the probability selection area 910 in FIG. 31) may be provided on the screen.

[0131] 28 is a diagram showing another example of the analysis screen 900. On the analysis screen 900, a scattergram 901 and a detailed display area 913 are displayed.

[0132] When the user performs an operation to select a dot on the scattergram 901, statistical information about the cell corresponding to the selected dot is displayed in the detailed display area 913. The detailed display area 913 has areas 913a to 913d. Area 913a displays the top two cell types with the highest probabilities for one or more cells plotted on the selected dot (hereinafter, the selected cells). Area 913b displays the cell type with the highest probability out of five white blood cell classifications for the selected cell. Area 913c displays the abnormal cell type with the highest probability for the selected cell. Area 913d displays all cell types that may be present for the selected cell, i.e., cell types with a probability greater than 0%.

[0133] According to the analysis screen 900 of FIG. 28, the user can refer to the detailed display area 913 to understand what cell type the selected cell may be classified into.

[0134] The probability of each cell type may be displayed adjacent to the cell type name displayed in areas 913a to 913d. If only one cell is selected, the probability information of the corresponding cell may be displayed. If multiple cells are selected, a representative value of the probabilities of the cell types of the multiple cells may be calculated and displayed. The representative value may be expressed, for example, as the mean, median, or mode. Furthermore, area 913a may display only the top cell type, or the top three or more cell types. Areas 913b and 913c may display the top two or more cell types. Area 913d may display cell types whose probabilities are greater than a predetermined threshold.

[0135] FIG. 29 is a flowchart showing a third example of the test result display process. In step S414, processor 3001 compares the number of cells corresponding to the dot selected on scattergram 901 with a threshold value. If the number of cells is equal to or less than the threshold value, processor 3001 displays the classification information for each cell individually in step S416. The individual display of the classification information for each cell is as described with reference to FIGS. 20 to 23. If the number of cells is greater than the threshold value, processor 3001 tallies the classification information and displays statistical information in step S415. The display of the statistical information is as described with reference to FIGS. 25 to 28.

[0136] As mentioned above, selecting multiple dots at once or setting a wide gate 901a may result in an excessive number of cells being displayed, and displaying the classification information for each cell individually may reduce user visibility. On the other hand, if the number of cells corresponding to the selected dots is small, displaying the classification information for each cell may be more suitable for the user's purposes. For example, if the user selects a location far from the center of a cell cluster or a location where cells are sparsely distributed, such as the dot selected on the scattergram 901 in Figure 28, the user may wish to analyze the cells individually rather than as a group. Therefore, in the example of Figure 29, when the number of selected dots is below a predetermined threshold, the classification information for the cells is displayed individually. When the number of selected dots is greater than the predetermined threshold, the classification information for the cells is displayed statistically instead of individually, automatically providing a highly visible screen for the user.

[0137] FIG. 30 is a diagram showing a fourth example of the test result display process. In the first to third examples of the test result process described above, an example was shown in which a dot of a target cell on the scattergram 901 is selected to display classification information of the cell corresponding to that dot. In the examples of FIGS. 31 to 38 described below, an analysis screen 900 including a graph is displayed (step S417), and dots of cells that satisfy the defined extraction conditions are highlighted (step S418). In step S418, instead of selecting dots on the scattergram 901, the user defines conditions for cells to be extracted, and cells that satisfy those conditions are extracted and displayed on the graph.

[0138] 31 is a diagram showing another example of the analysis screen 900. On the analysis screen 900, a scattergram 901, a cell type selection area 909, and a probability selection area 910 are displayed.

[0139] The cell type selection area 909 and the probability selection area 910 are configured as so-called pull-down menus. The cell type selection area 909 is configured to allow the user to select one cell type from all cell types corresponding to the measurement channel selected by the user (the WDF channel in FIG. 31). The probability selection area 910 displays probability values ​​selectable in 5% increments within the range of 0% to 100% as options. Specifically, the probability selection area 910 is configured to allow the user to select character strings that combine probability with an equal or inequality sign, such as "=100%", ">95%", ">90%", ..., ">5%, ">0%", "=0%", etc.

[0140] When the user selects a cell type of interest in cell type selection area 909 and a probability in probability selection area 910, the corresponding dots on scattergram 901 are highlighted. In the example shown in FIG. 31 , basophils are selected as the cell type, and “>85%” is selected as the probability. In this case, dots of cells with a basophil probability greater than 85% (dots in area 901b) are highlighted on scattergram 901. The highlighting may be performed by any display method that allows the extracted dots to be distinguished from other dots, such as by displaying the dots in a color different from other dots, highlighting the outline of the dots, or blinking the dots.

[0141] 31, the user can define the target cell type and probability as extraction conditions, and thereby grasp the position of cells that meet the conditions on the scattergram 901. This allows the user to grasp the position on the conventional scattergram 901 of cells classified by classification information (cell type and probability), and can use this information, for example, when judging the accuracy of classification based on classification information.

[0142] FIG. 32 shows another example of the analysis screen 900. Instead of the probability selection area 910, the analysis screen 900 in FIG. 32 includes a scale bar 910a for selecting a probability value and an extraction target selection area 910c. The scale bar 910a includes a pointer 910b that can slide left and right. The user can select any percentage within the range of 0% to 100% by moving the pointer 910b left and right. The extraction target selection area 910c displays a selectable inequality sign that can be combined with the percentage selected in the scale bar 910a. For example, as shown in FIG. 32, if basophils are selected as the cell type in the cell type selection area 909, the pointer 910b is placed at the 80% position in the scale bar 910a, and ">" is selected in the extraction target selection area 910c, the extraction condition becomes "probability of basophils > 80%." In the scattergram 901, dots of cells that satisfy the extraction condition "probability of basophils > 80%" are highlighted.

[0143] When the user moves the pointer 910b on the scale bar 910a, the range of percentage values ​​included in the extraction conditions is changed continuously (in stages). When the extraction conditions are changed, the dots extracted on the scattergram 901 are dynamically changed accordingly.

[0144] For example, if the pointer 910b is slid to the right from the state shown in FIG. 32, the extraction condition changes from "probability of basophils > 80%" to "probability of basophils > 85%." This change in extraction condition also dynamically changes the cell dots highlighted on the scattergram 901 from "probability of basophils > 80%" to "probability of basophils > 85%." In other words, the number of highlighted dots decreases. Conversely, if the pointer 910b is slid to the left, the extraction condition changes from "probability of basophils > 80%" to "probability of basophils > 75%." This change in extraction condition also dynamically changes the cell dots highlighted on the scattergram 901 from "probability of basophils > 80%" to "probability of basophils > 75%." In other words, the number of highlighted dots increases.

[0145] In this way, by sliding the pointer 910b on the scale bar 910a, the extraction conditions can be continuously changed, and accordingly the highlighting of the dots of the cells that match the extraction conditions displayed on the scattergram 901 is dynamically changed, allowing the user to easily determine where on the scattergram 901 the target cell is located.

[0146] FIG. 33 is a diagram showing another example of the analysis screen 900. FIG. 33 shows another example of setting extraction conditions. The analysis screen 900 in FIG. 33 includes condition setting areas 910d, 910e, and 910f for setting extraction condition 1, extraction condition 2, and extraction condition 3, respectively. The condition setting areas 910d, 910e, and 910f each include a pull-down menu corresponding to the cell type selection area 909 and probability selection area 910 shown in FIG. 31. Two operator setting areas 910g and 910h are provided between the condition setting areas 910d, 910e, and 910f. Each of the operator setting areas 910g and 910h is configured as a pull-down menu, allowing the user to set a logical operator for combining extraction conditions 1 to 3. Options for the logical operator include "AND," "OR," and "NOT." In the example in FIG. 33, all logical operators are "AND."

[0147] Extraction conditions 1 to 3 are defined in each of the condition setting areas 910d, 910e, and 910f, and logical operators are set in the operator setting areas 910g and 910h, thereby defining a composite condition in which extraction conditions 1 to 3 are linked by the logical operators. In the scattergram 901, dots of cells that satisfy the defined composite condition (dots in area 901b) are highlighted. As in the example of Figure 33, more detailed extraction conditions can be set by extracting cells using composite conditions that combine multiple extraction conditions, rather than just one extraction condition based on a one-to-one combination of cell type and probability.

[0148] Fig. 34 is a diagram showing another example of the analysis screen 900. The screen in Fig. 34 displays a scattergram 901, a first histogram 910i, a second histogram 910j, and a display area 910k.

[0149] 34, only a scattergram 901 is displayed by default on the analysis screen 900. The user can select a specific type of cell population on the displayed scattergram 901. For example, when the user places the cursor over a specific cell population on the scattergram 901, that cell population is selected.

[0150] FIG. 34 shows a state in which a lymphocyte population is selected. When a specific type of cell population is selected, a first histogram 910i representing the probability distribution of the selected cell population and a second histogram 910j representing the probability distribution of abnormal cells corresponding to the selected cell population are displayed on the analysis screen 900. The first histogram 910i and the second histogram 910j are similar to, for example, the histogram 905a in FIG. 25, and represent the number of cells on the probability axis. Abnormal cells corresponding to the selected cell population are abnormal cells that are morphologically similar to the selected cell population. Abnormal cells are determined based on a predetermined correspondence relationship depending on the selected cell population. For example, if lymphocytes are selected, the abnormal cells may be abnormal lymphocytes or blasts. For example, if monocytes are selected, the abnormal cells may be blasts. For example, if eosinophils or basophils are selected, the abnormal cells may be immature granulocytes.

[0151] In the example of Figure 34, when a specific cell population is selected, cells with a probability of being an abnormal cell corresponding to the selected cell population are extracted, and the distribution of these probabilities is displayed as a second histogram 910j. By comparing the probability distribution of the selected cell population with the probability distribution of abnormal cells, the user can understand how close the probability distribution of the selected cell population is to 100%. Alternatively, conversely, the user can statistically understand how many cells with a probability of being an abnormal cell are included in the selected cell population.

[0152] 34 includes a first histogram 910i and a second histogram 910j, as well as a display area 910k that displays an index based on statistical information. For example, the display area 910k displays "15% of cells in the lymphocyte population that have a probability of being an abnormal lymphocyte (previous value: 12%)." That is, the display area 910k displays, as an index, the percentage of cells with a corresponding probability of being an abnormal cell, with the selected cell population as the parent population.

[0153] In the example of Figure 34, the sample for the population is defined as "cells with a probability of being abnormal cells," but the definition of a sample is not limited to this. For example, it could also be "cells with the second highest probability of being abnormal cells." This makes it possible to extract only cells with a higher probability of being abnormal and calculate the index, which may reduce noise. Furthermore, the index is not limited to a quantitative index such as a ratio, but may be a qualitative index that displays a graded level according to the absolute number of cells with a probability of being abnormal cells.

[0154] In the example of Figure 34, the number or proportion of cells with a probability of being an abnormal cell corresponding to a specific cell population can be easily grasped. While Figures 25 to 28 described above have the advantage of allowing the user to arbitrarily specify a specific cell or cell population on the scattergram 901, it can be difficult to specify only a specific cell population on the scattergram 901. This can be particularly difficult when cells on the periphery of a cluster overlap with adjacent clusters or are distant from the cluster. In the example of Figure 34, simply by placing the cursor over a specific cell population on the scattergram 901, for example, over a lymphocyte population, a cell population containing lymphocytes as main cells can be selected. Furthermore, cells with a probability of being abnormal lymphocytes, which are abnormal cells corresponding to lymphocytes, are automatically extracted and a probability distribution and index are displayed, allowing the user to easily recognize the extent to which cells that are normal as main cells but have a probability of being abnormal cells are included.

[0155] Fig. 35 is a diagram showing another example of the analysis screen 900. On the screen of Fig. 35, a scattergram 901, a cell type selection area 909, and a histogram 911 are displayed.

[0156] 35 shows an example in which the WPC channel is designated as the measurement channel to be displayed. Therefore, the scattergram 901 is displayed based on the characteristic parameters of all cells corresponding to the measurement channel, and the cell type selection area 909 is configured to allow selection of all cell types associated with the WPC channel, such as mature white blood cells, abnormal lymphocytes, and blasts, as the cell types to be displayed.

[0157] When the user performs an operation to select a cell type in the cell type selection area 909, a histogram 911 is displayed based on the selected cell type. The histogram 911 is frequency distribution information obtained by aggregating the selected cell type by probability. In the histogram 911, the horizontal axis represents the probability, and the vertical axis represents the number of cells. When the user performs an operation to select a bar graph on the histogram 911, the corresponding dot (dot in area 901c) on the scattergram 901 is highlighted. In the example shown in FIG. 35, abnormal lymphocytes are selected as the cell type, and the bar graph with a probability of 70% is selected. In this case, the dots of cells with a probability of abnormal lymphocytes in the 70% range (cells distributed near area 901c) are highlighted on the scattergram 901.

[0158] 31 to 33, the analysis screen 900 in Fig. 35 allows the user to select a cell type and probability, thereby grasping the position of cells that meet the conditions on the scattergram 901. This allows the user to grasp the position on the conventional scattergram 901 of cells that can be classified by the classification information (cell type and probability).

[0159] 31 to 35, dots of cells that satisfy the extraction conditions, which are combinations of cell types and probabilities, are displayed on a scattergram 901. Instead of the scattergram 901, the number of cells that satisfy the extraction conditions may be displayed using a histogram, as shown in FIGS.

[0160] The analysis screen 900 in FIGS. 36 to 38 displays a probability selection area 910 and a number display area 912.

[0161] When the user selects a probability in the probability selection area 910 on the analysis screen 900 of FIG. 36, the number of cells corresponding to the probability selected in the probability selection area 910 among all cells corresponding to the target measurement channel is displayed in the number display area 912 for each cell type. The number display area 912 is provided with a descending order button 912a for sorting the cell types from top to bottom in descending order of number, and an ascending order button 912b for sorting the cell types from top to bottom in descending order of number. FIG. 36 shows the state in which the descending order button 912a has been operated, and FIG. 37 shows the state in which the ascending order button 912b has been operated. Furthermore, FIG. 38 shows the state in which ">20%" has been set as the probability condition on the screen of FIG. 37.

[0162] 36 to 38, after selecting a probability in the probability selection area 910, the user can set the display of the number display area 912 so that the numbers are arranged in descending order as shown in Fig. 36 by operating the descending order button 912a, and can set the display of the number display area 912 so that the numbers are arranged in ascending order as shown in Fig. 37 by operating the ascending order button 912b. This allows the user to smoothly check, for example, cell types with large numbers or cell types with small numbers among the cell types according to the probability of the selected condition.

[0163] By reselecting another probability in the probability selection area 910 from the state shown in Fig. 37, the user can reset the display in the number display area 912 according to the reselected probability, as shown in Fig. 38. In this way, by manipulating the probability selection area 910, the number of cell types based on different probability conditions can be smoothly confirmed.

[0164] 36 to 38 may also have a scattergram 901. In this case, when a bar graph in the number display area 912 is selected, the cells (points) corresponding to the selected bar graph may be highlighted in the scattergram 901. Also, a histogram 911 (see FIG. 35) may also be provided in FIGS. 36 to 38. In this case, when a bar graph in the number display area 912 is selected, the histogram 911 may display the probability distribution status for the cell type of the selected bar graph.

[0165] 39(a) and (b) are diagrams schematically showing other configurations of the scattergram 901. FIG.

[0166] In the scattergram 901 shown in Figure 39(a), distribution information of cells according to the probability of each cell type is superimposed along with plots of cells according to feature parameters (SSCP and SFLP in Figure 39(a)). In this case, the distribution information is composed of stripes connecting cells with the same probability in a contour line pattern, and distribution areas in which ranges of similar probability are represented by shades of gray. In Figure 39(a), the higher the probability, the darker the color of the distribution area. Furthermore, in Figure 39(a), cells (dots) based on the feature parameters are plotted on the scattergram 901, but as in Figure 39(b), dots of cells based on the feature parameters do not have to be plotted on the scattergram 901.

[0167] 39(a) and (b), the conventional scattergram 901 based on the characteristic parameters is displayed together with cell distribution information according to the probability of each cell type. This allows the user to visually understand how the probabilities of cell types are distributed in the conventional scattergram 901.

[0168] FIG. 40 is a flowchart showing a fifth example of the test result display process. In the first to fourth examples described above, the analysis screen 900 was displayed by double-clicking on the scattergram on the details screen 800. In the fifth example, when the "Research" tab is selected on the details screen 800 of FIG. 19, a research screen 900R is displayed instead of the analysis screen 900 (step S419). In this example, analysis results based on the main cell type are displayed on the sample list screen 700 and the details screen 800, and information or analysis result screens related to cell types other than the main cell type are displayed only on the research screen 900R. Then, based on the research flag, classification information for the selected cell type is read out, and a graph is displayed (step S420).

[0169] The screen in Fig. 41 is an example of a screen that is displayed when the research information button 921 is operated. The screen in Fig. 41 displays a cell type selection area 933 and a number display area 934. The cell type selection area 933 includes check boxes that allow the user to select the cell type related to the specified measurement channel.

[0170] When the user performs an operation to select a cell type in the cell type selection area 933, the count value of the cell types with a probability greater than 0% for the selected cell type is displayed in a bar graph for each cell type in the number display area 934.

[0171] In this way, after operating the research information button 921, the user can select a cell type in the cell type selection area 933 on the screen of Fig. 41, and then refer to the number display area 934 to check the detailed analysis results for the secondary cells. That is, as shown in Fig. 41, the research information display mode also displays the numbers of blast cells and abnormal lymphocytes, which normally cannot become main cells, allowing the user to perform a more detailed analysis based on the analysis results displayed in the research information display mode.

[0172] The screen in Fig. 42 is an example of a screen that is displayed when the research information button 921 is operated. A cell type selection area 933, a histogram 935, and a scattergram 936 are displayed on the screen in Fig. 42.

[0173] When the user performs an operation to select a cell type in the cell type selection area 933, a histogram 935 is displayed based on the selected cell type. The histogram 935 is frequency distribution information in which all cells corresponding to the measurement channel are tallied by the probability of the selected cell type. When the user selects a bar graph in the histogram 935, the corresponding cell (point) (dot in area 936a) on the scattergram 936 is highlighted. In the example shown in FIG. 42, a blast cell is selected as the cell type, and the bar graph for a probability of 50% is selected. In this case, cells with a blast probability in the 50% range (cells distributed near area 936a) are highlighted on the scattergram 936.

[0174] In this way, after operating the research information button 921, the user can check more detailed analysis results by selecting a cell type in the cell type selection area 933 on the screen of Figure 42 and selecting a bar graph in the histogram 935.

[0175] When multiple cell types are selected in the cell type selection area 933 in Fig. 42, graphs of the multiple cell types are displayed in histogram 935, as in the three-dimensional histogram 907 in Fig. 26. When multiple bar graphs are selected in histogram 935, all of the corresponding cells (points) on scattergram 936 are highlighted.

[0176] [7. Host computer transmission process] Next, a description will be given of the data (hereinafter referred to as "output data") sent from the processing unit 300 to the host computer 500. Figure 43 is a flowchart showing the host computer transmission subroutine.

[0177] The processor 3001 of the processing unit 300 receives a validation operation from the user, and generates output data to be sent to the host computer 500 in response to the validation operation (step S141).

[0178] As described with reference to Figures 18 and 19, the test result screens, namely the sample list screen 700 and the details screen 800, display a validate icon 710d that is displayed in common between the screens. The user checks the test result screen described above and verifies the test results. If the user determines that the results can be reported to the host computer 500 as a result of the verification, the user clicks the validate icon 710d. If the user does not perform validation on the target sample and sets up a retest for the target subject, the user operates the measurement registration icon 710c to create a test order for the retest.

[0179] When the validation operation is performed, the processor 3001 generates output data to be transmitted to the host computer 500 based on the analysis result data stored in the storage unit 3004 (step S141). Then, the processor 3001 transmits the generated output data to the host computer 500 (step S142).

[0180] 44(a) to 45(b) are diagrams showing the structure of output data generated by the processor 3001 and transmitted to the host computer 500. The output data may be formed by removing classification information corresponding to a predetermined cell type from the test result data, and may be formed, for example, as shown in any one of FIGS. 44(a) to 45(b).

[0181] The output data shown in Figure 44(a) is composed of the specimen ID, measurement result, measurement channel, cell ID, identification information regarding the main cell type, and probability regarding the main cell type. In other words, the output data shown in Figure 44(a) is the test result data shown in Figures 12 to 16, with data other than the output data shown in Figure 44(a) removed.

[0182] The output data shown in Figure 44(b) is composed of the specimen ID, measurement result, measurement channel, cell ID, identification information for the top two cell types with the highest probabilities, and the probabilities for the top two cell types with the highest probabilities. That is, the output data shown in Figure 44(b) is the output data shown in Figure 44(a) to which identification information and probability for the cell type with the second highest probability have been added. Note that the output data shown in Figure 44(b) is not limited to having the top two cell types and probabilities, and may also have cell types and probabilities from the top down to a predetermined rank.

[0183] The output data shown in Figure 45(a) is composed of a sample ID and a measurement result. That is, the output data shown in Figure 45(a) is the test result data shown in Figures 12 to 16 with all data other than the output data shown in Figure 45(a) removed. That is, the output data does not include any information regarding probability.

[0184] The output data shown in FIG. 45(b) is composed of a specimen ID, measurement results, and abnormal cell detection information. That is, the output data shown in FIG. 45(b) is the same as the output data shown in FIG. 45(a), but with abnormal cell detection information added to the measurement results. If a predetermined number of abnormal cells (e.g., abnormal lymphocytes or blasts) with a probability greater than 0% are present in the specimen, the fact that the abnormal cells were detected is added to the measurement results as abnormal cell detection information when the test result data is generated. The abnormal cell detection information is, for example, a string of characters such as "Abnormal lymphocytes detected," "Abn LYMPH?", "Blasts detected," or "Blast?". The abnormal cell detection information may also be the cell type and number of detected abnormal cells. The abnormal cell detection information is displayed, for example, in the measurement result display area 932 of FIG. 18.

[0185] As described above, output data containing the counting results and excluding at least some of the information related to the relationships (such as cell types and probabilities) is transmitted to the host computer 500. As described above, when probabilities associated with multiple cell types are obtained for each cell, the volume of test result data becomes enormous. Therefore, if all of the test result data were transmitted to the host computer 500, the communication load due to data transmission would increase. In contrast, according to the output data shown in Figures 44(a) to 45(b), only the data necessary for management by the host computer 500 is transmitted, rather than transmitting all of the vast amount of test result data. This allows for both appropriate data management in the host computer 500 and efficient data transmission.

[0186] The output data shown in FIGS. 44(a) to 45(b) may include characteristic parameters and scattergrams.

[0187] Here, the processor 3001 generates the output data so that it has one of the configurations shown in Figures 44(a) to 45(b). However, the configuration of the output data may be set in accordance with an instruction from the user.

[0188] 46(a) and 46(b) are diagrams each showing a schematic configuration of a screen for selecting data to be included in output data.

[0189] The screen in Figure 46(a) displays a cell type selection area 951 and an OK button 952. The cell type selection area 951 is provided with check boxes for selecting a cell type. When the user performs an operation to select a cell type in the cell type selection area 951 and operates the OK button 952, the processor 3001 of the processing unit 300 stores the setting contents in the memory section 3004. From the next time that the processor 3001 generates output data, it will include the set cell type and the probability corresponding to the set cell type in the output data.

[0190] The screen in Figure 46(b) displays a cell range selection area 961 and an OK button 962. The cell range selection area 961 has check boxes for selecting the range of cell types. In the cell range selection area 961, it is possible to select the range of cell types, such as "main cell type only," "sub-cell types other than main cell type," "all cell types," and "do not send." When "main cell type only" is selected, of all the identification information and probabilities, only the identification information and probabilities of the main cell type are included in the output data. When "sub-cell types other than main cell type" is selected, of all the identification information and probabilities, only the identification information and probabilities of all sub-cell types other than the main cell type are included in the output data. When "all cell types" is selected, all the identification information and probabilities are included in the output data. When "do not send" is selected, all the identification information and probabilities are not included in the output data.

[0191] When the user performs an operation to select a range of cell types in the cell range selection area 961 and operates the OK button 962, the processor 3001 stores the setting contents in the storage unit 3004. From the next time onwards, when the processor 3001 generates output data, it generates the output data based on the set range of cell types.

[0192] [8. Processor and parallel processing processor configuration] (Configuration example 1) Next, the configuration of the parallel processing processor 4833 mounted on the measurement unit 400 will be described in detail with reference again to Fig. 2. Below, configuration example 1 will be described regarding the processor 4831 and the parallel processing processor 4833.

[0193] As described with reference to FIG. 2 , the processor 4831 uses the parallel processing processor 4833 to perform analysis processing of waveform data included in the generated digital signal in accordance with the deep learning algorithm 60. That is, the processor 4831 is programmed to perform analysis processing of waveform data included in the digital signal in accordance with the deep learning algorithm 60. Analysis software 4835a for analyzing cell data based on the deep learning algorithm 60 may be stored in the storage unit 4835. In this case, the processor 4831 executes the analysis software 4835a stored in the storage unit 4835 to perform data analysis processing based on the deep learning algorithm 60. The processor 4831 is, for example, a CPU (Central Processing Unit). The processor 4831 may be, for example, a Core i9, Core i7, or Core i5 manufactured by Intel Corporation, or a Ryzen 9, Ryzen 7, Ryzen 5, or Ryzen 3 manufactured by AMD.

[0194] The processor 4831 controls the parallel processing processor 4833. The parallel processing processor 4833 executes parallel processing relating to, for example, matrix operations, according to the control of the processor 4831. In other words, the processor 4831 is the master processor of the parallel processing processor 4833, and the parallel processing processor 4833 is a slave processor of the processor 4831. The processor 4831 is also called a host processor or a main processor.

[0195] The parallel processing processor 4833 executes multiple arithmetic processes in parallel, which are at least a part of the processing related to the analysis of waveform data. The parallel processing processor 4833 is, for example, a graphics processing unit (GPU), a field programmable gate array (FPGA), or an application specific integrated circuit (ASIC). When the parallel processing processor 4833 is an FPGA, the parallel processing processor 4833 may be pre-programmed with arithmetic processes related to the trained deep learning algorithm 60, for example. When the parallel processing processor 4833 is an ASIC, the parallel processing processor 4833 may be pre-programmed with a circuit for executing arithmetic processes related to the trained deep learning algorithm 60, for example, or may have a programmable module built in addition to such a built-in circuit. The parallel processing processor 4833 may be, for example, NVIDIA GeForce, Quadro, TITAN, Jetson, or the like. For the Jetson series, for example, Jetson Nano, Jetson Tx2, Jetson Xavier, or Jetson AGX Xavier may be used.

[0196] The processor 4831 executes, for example, calculations related to the control of the measurement unit 400. The processor 4831 executes, for example, calculations related to control signals transmitted and received between the device mechanism 430, the sample preparation unit 440, and the specimen aspirating unit 450. The processor 4831 executes, for example, calculations related to the transmission and reception of information to and from the processing unit 300. The processor 4831 executes, for example, processes related to reading program data from the storage unit 4835, loading the program into the RAM 4834, and transmitting and receiving data to and from the RAM 4834. The above-described processes executed by the processor 4831 are required to be executed in a predetermined order, for example. For example, if processes A, B, and C are required to control the device mechanism 430, the sample preparation unit 440, and the specimen aspirating unit 450, they may be required to be executed in the order B, A, and C. The processor 4831 often executes sequential processes that depend on such an order, and therefore, increasing the number of arithmetic units (sometimes called "processor cores," "cores," etc.) does not necessarily increase the processing speed.

[0197] On the other hand, the parallel processing processor 4833 executes routine, large-volume computations, such as operations on matrix data containing a large number of elements. In this embodiment, the parallel processing processor 4833 executes parallel processing, which parallelizes at least a portion of the processing for analyzing waveform data according to the deep learning algorithm 60. The deep learning algorithm 60 may include, for example, a large number of matrix operations. The deep learning algorithm 60 may include, for example, at least 100 matrix operations, or may include at least 1,000 matrix operations. The parallel processing processor 4833 has multiple arithmetic units, each of which can simultaneously execute a matrix operation. In other words, the parallel processing processor 4833 can execute matrix operations in parallel by each of the multiple arithmetic units as parallel processing. For example, the matrix operation included in the deep learning algorithm 60 can be divided into multiple arithmetic operations that are independent of each other in order. The divided arithmetic operations can then be executed in parallel by each of the multiple arithmetic units. These arithmetic units may be referred to as "processor cores," "cores," or the like.

[0198] By performing such parallel processing, it is possible to speed up the calculation processing of the measurement unit 400 as a whole. Processing such as matrix operations included in the deep learning algorithm 60 is sometimes called, for example, "Single Instruction Multiple Data" (SIMD). The parallel processing processor 4833 is suitable for such SIMD operations, for example. Such a parallel processing processor 4833 is sometimes called a vector processor.

[0199] As described above, the processor 4831 is suitable for executing a variety of complex processes. On the other hand, the parallel processing processor 4833 is suitable for executing a large amount of standardized processes in parallel. By executing a large amount of standardized processes in parallel, the TAT (Turn Around Time) required for calculation processing is shortened.

[0200] Note that the target of parallel processing executed by the parallel processing processor 4833 is not limited to matrix operations. For example, when the parallel processing processor 4833 executes learning processing according to the deep learning algorithm 60, differential operations and the like related to the learning processing can be targets of parallel processing.

[0201] The number of arithmetic units of the processor 4831 is, for example, a dual-core (number of cores: 2), a quad-core (number of cores: 4), or an octa-core (number of cores: 8). On the other hand, the parallel processing processor 4833 has, for example, at least 10 arithmetic units (number of cores: 10) and can execute 10 matrix operations in parallel. Some parallel processing processors 4833 have, for example, several tens of arithmetic units. Other parallel processing processors 4833 have, for example, at least 100 arithmetic units (number of cores: 100) and can execute 100 matrix operations in parallel. Some parallel processing processors 4833 have, for example, several hundred arithmetic units. Other parallel processing processors 4833 have, for example, at least 1,000 arithmetic units (number of cores: 1,000) and can execute 1,000 matrix operations in parallel. The parallel processing processor 4833 may have, for example, several thousand arithmetic units.

[0202] FIG. 47 shows an example configuration of the parallel processing processor 4833. The parallel processing processor 4833 includes a plurality of arithmetic units 4836 and a RAM 4837. Each of the arithmetic units 4836 executes arithmetic processing of matrix data in parallel. The RAM 4837 stores data related to the arithmetic processing executed by the arithmetic units 4836. The RAM 4837 is a memory having a capacity of at least 1 gigabyte. The RAM 4837 may also have a capacity of 2 gigabytes, 4 gigabytes, 6 gigabytes, 8 gigabytes, 10 gigabytes, or more. The arithmetic units 4836 obtain data from the RAM 4837 and execute arithmetic processing. The arithmetic units 4836 may be referred to as a "processor core," a "core," or the like.

[0203] 48 to 50 show an example of mounting the parallel processing processor 4833 in the measuring unit 400. FIGS. 48 to 50 show an example in which the parallel processing processor 4833 is mounted in the cell analyzer 100 by being incorporated inside the measuring unit 400. FIGS. 48 and 49 show an example in which the processor 4831 and the parallel processing processor 4833 are provided as separate units. As shown in FIG. 48, the processor 4831 is mounted on, for example, a substrate 4838. The parallel processing processor 4833 is mounted on, for example, a graphic board 4830, which is connected to the substrate 4838 via a connector 4839. The processor 4831 is connected to the parallel processing processor 4833 via a bus 485. As shown in FIG. 49, the parallel processing processor 4833 may be mounted directly on the substrate 4838 and connected to the processor 4831 via the bus 485. 50 shows an example of a mounting arrangement in which a processor 4831 and a parallel processing processor 4833 are integrally provided. As shown in FIG. 50, the parallel processing processor 4833 may be built into the processor 4831 mounted on a substrate 4838, for example.

[0204] FIG. 51 shows another example of mounting the parallel processing processor 4833 on the measuring unit 400. FIG. 51 shows an example in which the parallel processing processor 4833 is mounted on the measuring unit 400 by an external device 4800 connected to the measuring unit 400. The parallel processing processor 4833 is mounted on the external device 4800, which is, for example, a USB (Universal Serial Bus) device, and the USB device is connected to the bus 485 via an interface unit 487, thereby mounting the parallel processing processor 4833 on the cellular analyzer 100. The USB device may be, for example, a small device such as a USB dongle. The interface unit 487 is, for example, a USB interface having a transfer speed of several hundred Mbps, more preferably a USB interface having a transfer speed of several Gbps to several tens of Gbps or more. For example, the Neural Compute Stick 2 manufactured by Intel Corporation may be used as the external device 4800 on which the parallel processing processor 4833 is mounted.

[0205] A plurality of parallel processing processors 4833 may be mounted on the cell analyzer 100 by connecting a plurality of USB devices each equipped with a parallel processing processor 4833 to the interface unit 487. Since the parallel processing processor 4833 mounted on one USB device may have a smaller number of arithmetic units 4836 than a GPU or the like, the number of cores can be increased by adding a plurality of USB devices connected to the measurement unit 400.

[0206] 52, 53(a), 53(b), and 54 show an overview of the calculation process executed by the parallel processing processor 4833 under the control of the analysis software 4835a running on the processor 4831.

[0207] FIG. 52 shows an example of the configuration of the parallel processor 4833 that executes the arithmetic processing.

[0208] The parallel processing processor 4833 has multiple arithmetic units 4836 and a RAM 4837. The processor 4831, which executes the analysis software 4835a, commands the parallel processing processor 4833 to cause the parallel processing processor 4833 to execute at least a portion of the arithmetic processing required to analyze waveform data using the deep learning algorithm 60. The processor 4831 commands the parallel processing processor 4833 to execute arithmetic processing related to waveform data analysis based on the deep learning algorithm. All or at least a portion of the waveform data corresponding to the signal detected by the FCM detection unit 410 is stored in the RAM 4834. The data stored in the RAM 4834 is transferred to the RAM 4837 of the parallel processing processor 4833. The data stored in the RAM 4834 is transferred to the RAM 4837, for example, using a direct memory access (DMA) method. Each of the multiple arithmetic units 4836 of the parallel processing processor 4833 executes arithmetic processing on the data stored in the RAM 4837 in parallel. Each of the multiple arithmetic units 4836 obtains necessary data from the RAM 4837 and executes arithmetic processing. Data corresponding to the arithmetic results is stored in the RAM 4837 of the parallel processor 4833. The data corresponding to the arithmetic results is transferred from the RAM 4837 to the RAM 4834, for example, by DMA.

[0209] 53(a) and (b) show an overview of the matrix operations executed by the parallel processing processor 4833.

[0210] When analyzing waveform data according to the deep learning algorithm 60, matrix multiplication (matrix operation) is performed. The parallel processing processor 4833, for example, executes multiple arithmetic processes related to matrix operations in parallel. FIG. 53(a) shows a formula for calculating matrix multiplication. In the formula shown in (a), matrix c is obtained by multiplying matrix a with n rows and n columns by matrix b with n rows and n columns. As shown in FIG. 53(a), the formula is written using a multi-layer loop syntax. FIG. 53(b) shows an example of arithmetic processes executed in parallel by the parallel processing processor 4833. The formula shown in FIG. 53(a) can be divided into n×n arithmetic processes, where n is the number of combinations of loop variable i in the first layer and loop variable j in the second layer. Each of the divided arithmetic processes is independent of each other and can therefore be executed in parallel.

[0211] FIG. 54 is a conceptual diagram showing that the plurality of arithmetic processes illustrated in FIG. 53(b) are executed in parallel by the parallel processing processor 4833.

[0212] 54, each of the multiple arithmetic processes is assigned to one of the multiple arithmetic units 4836 included in the parallel processing processor 4833. Each of the arithmetic units 4836 executes the assigned arithmetic processes in parallel with each other. In other words, each of the arithmetic units 4836 executes the divided arithmetic processes simultaneously.

[0213] 53(a), (b) and 54, information regarding the probability that a cell corresponding to waveform data belongs to each of a plurality of cell types is obtained. Based on the results of the calculations, the processor 4831 executing the analysis software 4835a performs an analysis regarding the cell type of the cell corresponding to the waveform data. The calculation results are stored in the RAM 4837 of the parallel processor 4833 and transferred from the RAM 4837 to the RAM 4834. The processor 4831 transmits the results of the analysis based on the calculation results stored in the RAM 4834 to the processing unit 300 via the bus 485 and the interface unit 489.

[0214] The calculation of the probability that a cell belongs to each of the plurality of cell types may be performed by a processor other than the parallel processing processor 4833. For example, the calculation results by the parallel processing processor 4833 may be transferred from the RAM 4837 to the RAM 4834, and the processor 4831 may calculate information regarding the probability that a cell corresponding to each waveform data belongs to each of the plurality of cell types based on the calculation results read from the RAM 4834. Alternatively, the calculation results by the parallel processing processor 4833 may be transferred from the RAM 4837 to the processing unit 300, and a processor mounted on the processing unit 300 may calculate information regarding the probability that a cell corresponding to each waveform data belongs to each of the plurality of cell types.

[0215] In this embodiment, the processing shown in Figures 53(a), (b) and 54 is applied to, for example, calculation processing (also called filtering processing) related to a convolution layer in the deep learning algorithm 60.

[0216] Figures 55(a) and (b) show an overview of the computational processing for the convolutional layer.

[0217] FIG. 55(a) shows an example of waveform data of forward scattered light (FSC) as waveform data input to the deep learning algorithm 60. The waveform data in this embodiment is one-dimensional matrix data, as described with reference to FIG. 5. More simply, the waveform data is an array in which elements are arranged in a row. For ease of explanation, the number of elements in the waveform data is assumed to be n (n is an integer equal to or greater than 1). FIG. 55(a) shows multiple filters. The filters are generated by the learning process of the deep learning algorithm 60. Each of the multiple filters is one-dimensional matrix data representing the characteristics of the waveform data. The filter shown in FIG. 55(a) is matrix data with one row and three columns, but the number of columns is not limited to three. By performing a matrix operation on the waveform data input to the deep learning algorithm 60 and each filter, characteristics corresponding to the cell type related to the waveform data are calculated. FIG. 55(b) shows an overview of the matrix operation between the waveform data and the filters. As shown in FIG. 55(b), the matrix operation is performed while shifting each filter by one position relative to each element of the waveform data. The matrix operation is calculated according to the following equation 1.

[0218]

number

[0219] In Equation 1, the subscripts of x are variables indicating the row and column numbers of the waveform data. The subscripts of h are variables indicating the row and column numbers of the filter. In the example shown in Figures 55(a) and 55(b), the waveform data is one-dimensional matrix data, and the filter is matrix data with 1 row and 3 columns, so L = 1, M = 3, p = 0, q = 0, 1, 2, i = 0, j = 0, 1, ..., n-1.

[0220] The parallel processing processor 4833 executes the matrix operation expressed by Equation 1 in parallel using each of the multiple calculation units 4836. Classification information regarding the cell type of each cell is generated based on the calculation processing executed by the parallel processing processor 4833. The generated classification information is transmitted to the processing unit 300, which uses the classification information to generate and display test results for the specimen.

[0221] Next, with reference to FIG. 56, the cell classification process in step S13 in the flowchart of FIG. 10 will be described.

[0222] The cell classification process in step S13 is a process performed by the processor 4831 in response to the operation of the analysis software 4835a. The processor 4831 transfers the digital signal taken into the RAM 4834 in step S13 to the parallel processing processor 4833 (step S101). The processor 4831 transfers the digital signal from the RAM 4834 to the RAM 4837 by DMA transfer, as shown in Fig. 52. The processor 4831 controls, for example, the bus controller 4850 to cause the digital signal to be DMA transferred from the RAM 4834 to the RAM 4837.

[0223] The processor 4831 instructs the parallel processing processor 4833 to execute parallel processing on the waveform data included in the digital signal (step S102). The processor 4831 instructs the parallel processing processor 4833 to execute parallel processing, for example, by calling a kernel function of the parallel processing processor 4833. The processing executed by the parallel processing processor 4833 will be described later using the flowchart illustrated in FIG. 57. The processor 4831 instructs the parallel processing processor 4833 to execute a matrix operation related to the deep learning algorithm 60, for example. The digital signal is decomposed into multiple waveform data, which are sequentially input to the deep learning algorithm 60. The indexes corresponding to each cell included in the digital signal are not input to the deep learning algorithm 60. The waveform data input to the deep learning algorithm 60 is operated on by the parallel processing processor 4833.

[0224] The processor 4831 receives the results of the calculations executed by the parallel processing processor 4833 (step S103). The results of the calculations are transferred by DMA from the RAM 4837 to the RAM 4834, for example, as shown in FIG.

[0225] The processor 4831 generates an analysis result of the cell type of each measured cell based on the calculation result by the parallel processing processor 4833 (step S104).

[0226] FIG. 57 shows an example of the operation of the parallel processing processor 4833 which is executed based on instructions from the processor 4831 in response to the operation of the analysis software 4835a.

[0227] The processor 4831 executing the analysis software 4835a causes the parallel processing processor 4833 to assign arithmetic operations to the arithmetic units 4836 (step S110). The processor 4831 causes the parallel processing processor 4833 to assign arithmetic operations to the arithmetic units 4836, for example, by calling a kernel function of the parallel processing processor 4833. As shown in FIG. 54 , for example, a matrix operation related to the deep learning algorithm 60 is divided into multiple arithmetic operations, and each of the divided arithmetic operations is assigned to the arithmetic units 4836. Multiple waveform data are input sequentially to the deep learning algorithm 60. A matrix operation corresponding to the waveform data is divided into multiple arithmetic operations, and the divided arithmetic operations are assigned to the arithmetic units 4836.

[0228] Each arithmetic process is performed in parallel by a plurality of arithmetic units 4836 (step S111). The arithmetic process is executed on a plurality of waveform data.

[0229] The operation results generated by the parallel processing by the multiple operation units 4836 are transferred from the RAM 4837 to the RAM 4834 (step S112). For example, the operation results are transferred from the RAM 4837 to the RAM 4834 by DMA.

[0230] (Configuration example 2) 58 and 59, another configuration example of the cell analyzer 100 configured by the measuring unit 400 and the processing unit 300 will be described. In this configuration example 2, the parallel processor is provided in the processing unit 300.

[0231] FIG. 58 shows a block diagram of the measurement unit 400 in the second configuration example.

[0232] 58 does not include an A / D conversion unit 482, a processor 4831, a RAM 4834, a storage unit 4835, or a parallel processing processor 4833, and has the same configuration and functions as the measurement unit 400 of configuration example 1 described in Figures 1 to 57 and the related descriptions, except that it is provided with a connection port 4201. A connection cable 4202 is connected to the connection port 4201.

[0233] FIG. 59 shows a block diagram of the processing unit 300 in the second configuration example.

[0234] 59, the processing unit 300 includes a processor 3001, a parallel processing processor 3002, a memory unit 3004, a RAM 3005, an interface unit 3006, an A / D conversion unit 3008, a bus controller 4850, and an interface unit 3009, which are connected to a bus 3003. In other words, in the example of FIG. 59, the parallel processing processor 3002 is incorporated into the processing unit 300 and is installed in the cell analyzer 100.

[0235] The bus 3003 is, for example, a transmission line having a data transfer rate of several hundred MB / s or more. The bus 3003 may also be a transmission line having a data transfer rate of 1 GB / s or more. The bus 3003 transfers data based on, for example, the PCI-Express or PCI-X standard. The configurations of the processor 3001, parallel processing processor 3002, storage unit 3004, and RAM 3005, and the processes executed therein, are similar to the configurations and processes of the processor 4831, parallel processing processor 4833, storage unit 4835, and RAM 4834 described above in FIGS. 47 to 57. The A / D conversion unit 3008 samples the analog signal output from the measurement unit 400 as described above, and generates a digital signal containing waveform data of the cell. The method of generating the digital signal is as described above.

[0236] 58 and 59, the connection cable 4202 includes, for example, transmission paths whose number corresponds to the types of analog signals transmitted from the measurement unit 400 to the processing unit 300. For example, the connection cable 4202 is configured with a twisted pair cable, and has a number of pairs of wires whose number corresponds to the types of analog signals transmitted to the processing unit 300. The transmission path from the connection port 3007 to the A / D conversion unit 3008 may also have a number of wires whose number corresponds to the types of analog signals transmitted to the processing unit 300. In the transmission path from the connection port 3007 to the A / D conversion unit 3008, for example, the analog signals are transmitted as differential signals.

[0237] As shown in FIG. 60, the processor 3001 and the parallel processing processor 3002 have the same configuration and functions as the above-described processor 4831 and parallel processing processor 4833. The parallel processing processor 3002 includes a plurality of arithmetic units 3200 and a RAM 3201. Analysis software 3100 that analyzes the cell type of measured cells runs on the processor 3001. Note that the parallel processing processor 3002 does not need to be directly connected to the bus 3003. For example, it may be implemented in a USB device, and the USB device may be connected to the bus 3003 via an interface unit (not shown), thereby being installed in the cell analysis device 100 as part of the processing unit 300. The USB device may be, for example, a small device such as a USB dongle.

[0238] 58 and 59, the processing unit 300 is connected to the interface section 489 of the measurement unit 400 via the interface section 3006. The processing unit 300 transmits control signals for the device mechanism section 430 and the sample preparation section 440 to the measurement unit 400 via the interface section 3006. The interface section 3006 is, for example, a USB interface.

[0239] 58 and 59 , the processing unit 300 is connected to a connection port 4201 of the measurement unit 400 via an interface unit 3006, a connection port 3007 connected to an A / D conversion unit 3008, and a connection cable 4202 connected to the connection port 3007. The connection port 4201 is connected to an analog processing unit 420. As described above, the analog signal output from the connection port 4201 to the processing unit 300 is a signal obtained by processing the output of the FCM detection unit 410 of the measurement unit 400 by the analog processing unit 420. The analog processing unit 420 performs processing, including noise removal, on the analog signal input from the FCM detection unit 410. The analog signal processed by the analog processing unit 420 is transmitted to the processing unit 300 via the connection port 4201 and the connection cable 4202. The connection cable 4202 is configured to be, for example, one meter or less in length to reduce noise during signal transmission. The analog signal is transmitted, for example as a differential signal, to the processing unit 300 via a connecting cable 4202. In this case, the connecting cable 4202 is preferably a twisted pair cable.

[0240] The processing unit 300 may include a plurality of connection ports 3007. The processing unit 300 may acquire analog signals from a plurality of measurement units 400 via the plurality of connection ports 3007.

[0241] The analog signal transmitted from the measurement unit 400 via the connection cable 4202 is converted into a digital signal by the A / D converter 3008 of the processing unit 300. The A / D converter 3008 samples the transmitted analog signal at a predetermined sampling rate (e.g., sampling of 1024 points at 10 nanosecond intervals, sampling of 128 points at 80 nanosecond intervals, or sampling of 64 points at 160 nanosecond intervals), as described with reference to FIG. 5, to generate waveform data for each cell. The waveform data is stored in the memory unit 3004 or RAM 3005 via the bus 3003. The waveform data is transferred to the RAM 3005 by, for example, DMA. The processor 3001 and the parallel processing processor 3002 perform arithmetic processing on the waveform data stored in the memory unit 3004 or RAM 3005.

[0242] Analysis software 3100 running on processor 3001 has the same functions as analysis software 4835a shown in Fig. 52. By executing analysis software 3100, processor 3001 generates classification information regarding the cell type of the measured cells through operations similar to those shown in Fig. 52, Fig. 53(a) and (b), Fig. 54, Fig. 56, Fig. 57 and their related descriptions.

[0243] In the case of the cell analyzer 100 of Configuration Example 2 shown in Figures 58 and 59, the generation of waveform data and feature parameters in step S12 of the flowchart shown in Figure 10 and the cell classification using a deep learning algorithm in step S13 are performed in the processing unit 300. Step S14 (transmission of classification information) is omitted. The processing in Figures 56 and 57 is performed by the processor 3001 and parallel processing processor 3002 of the processing unit 300.

[0244] (Configuration example 3) 61 and 62, another configuration example of a cell analyzer 100 configured by a measurement unit 400 and a processing unit 300 will be described. In this configuration example 3, too, the parallel processing processor 3002 is mounted on the cell analyzer 100 in a form that is incorporated inside the processing unit 300.

[0245] FIG. 61 shows a block diagram of the measurement unit 400 in the third configuration example.

[0246] The measurement unit 400 shown in Figure 61 does not have a processor 4831, RAM 4834, memory unit 4835, or parallel processing processor 4833, and has the same configuration and function as the measurement unit 400 of configuration example 1 described in Figures 2, 47, and their related descriptions, except that it is provided with an interface unit 4851 and a transmission path 4852 for transmitting the digital signal generated by the A / D conversion unit 482 to the processing unit 300.

[0247] The interface unit 4851 is, for example, an interface serving as a dedicated line having a communication bandwidth of 1 gigabit / second or more. For example, the interface unit 4851 is an interface conforming to Gigabit Ethernet, USB 3.0, or Thunderbolt 3. If the interface unit 4851 is Gigabit Ethernet, the transmission path 4852 is, for example, a LAN cable. If the interface unit 4851 is USB 3.0, the transmission path 4852 is, for example, a USB cable conforming to USB 3.0. The transmission path 4852 is, for example, a dedicated transmission path for transmitting digital signals between the measurement unit 400 and the processing unit 300.

[0248] FIG. 62 shows a block diagram of the processing unit 300 in the third configuration example.

[0249] 62 has the same configuration and functions as the processing unit 300 of configuration example 2 described in FIG. 59 and its related description, except that it does not include an A / D conversion unit 3008 and a connection port 3007 and includes an interface unit 3010. The processing unit 300 may be connected to a plurality of measurement units 400 via a plurality of interface units 3010 and a plurality of interface units 3006.

[0250] The processor 3001 and the parallel processing processor 3002 have the same configuration and functions as the processor 3001 and the parallel processing processor 3002 described in Figure 60 and its related description. In Figure 62, the parallel processing processor 3002 does not necessarily need to be directly connected to the bus 3003, but may be implemented in, for example, a USB device, and this USB device may be connected to the bus 3003 via an interface unit (not shown). This USB device may be, for example, a small device such as a USB dongle.

[0251] Analysis software 3100 running on processor 3001 has the same functions as analysis software 4835a shown in Fig. 52. Analysis software 3100 analyzes the cell type of the measured cells by operations similar to those shown in Fig. 52, Fig. 53(a) and (b), Fig. 54, Fig. 56, Fig. 57 and their related descriptions.

[0252] In the case of the cell analyzer 100 of Configuration Example 3 shown in Figures 61 and 62, step S13 (cell classification) in the flowchart shown in Figure 10 is performed in the processing unit 300. Step S14 (transmission of classification information) is omitted. The processing in Figures 56 and 57 is performed by the processor 3001 and parallel processing processor 3002 of the processing unit 300.

[0253] In the configurations of FIGS. 61 and 62 , analog signals (forward scattered light signal, side scattered light signal, side fluorescent light signal) of cells generated in the FCM detection unit 410 are converted into digital signals in the A / D conversion unit 482 in the measuring unit 400. The digital signals are sent to the processing unit 300 via the interface unit 484, bus 485, interface unit 4851, and transmission path 4852. As described above, the transmission path 4852 is a transmission path dedicated to transmitting digital signals between the measuring unit 400 and the processing unit 300. For example, the measuring unit 400 and the processing unit 300 are connected one-to-one via the transmission path 4852. In other words, the transmission path 4852 is a transmission path that does not involve transmission of data related to devices other than the components (e.g., the measuring unit 400 and the processing unit 300) that make up the cell analysis device 100. The transmission path 4852 is a transmission path separate from, for example, an intranet or the Internet. This makes it possible to avoid bottlenecks in the communication speed of digital signal transmission even when digital signals generated by A / D conversion within the measurement unit 400 are transmitted to the processing unit 300.

[0254] (Configuration example 4) A fourth configuration example of the cell analyzer 100 will be described with reference to FIGS. 63, 64, 65, 66, and 67.

[0255] In this configuration example 4, as illustrated in Figure 63, an analysis unit 600 is provided between the measurement unit 400 and the processing unit 300. That is, in the configurations of Figures 63, 64, 65, 66, and 67, the cell analyzer 100 is configured to include the measurement unit 400, the processing unit 300, and the analysis unit 600. The analysis unit 600 analyzes the cell type of the measured cells. As will be described later, in this configuration example, the parallel processing processor 6002 is installed in the cell analyzer 100 by being incorporated into the analysis unit 600.

[0256] FIG. 64 illustrates the configuration of the measurement unit 400 in the fourth configuration example.

[0257] The configuration of the measurement unit 400 illustrated in FIG. 64 has the same configuration and functions as the measurement unit 400 of Configuration Example 3 described in FIG. 61 and its related description. An analysis unit 600 is provided between the measurement unit 400 and the processing unit 300. The analysis unit 600 may be connected to multiple measurement units 400. The analysis unit 600 may be connected to multiple processing units 300. The interface unit 4851 is, for example, an interface having a communication bandwidth of 1 gigabit / second or more. For example, the interface unit 4851 is an interface compliant with Gigabit Ethernet, USB 3.0, or Thunderbolt 3. If the interface unit 4851 is Gigabit Ethernet, the transmission path 4852 is, for example, a LAN cable. If the interface unit 4851 is USB 3.0, the transmission path 4852 is a USB cable compliant with USB 3.0. As described above, the transmission path 4852 is a dedicated transmission path for transmitting digital signals between the measurement unit 400 and the processing unit 300. For example, the measurement unit 400 and the processing unit 300 are connected one-to-one via a transmission path 4852 .

[0258] FIG. 65 shows an example of the configuration of the analysis unit 600.

[0259] The analysis unit 600 includes, for example, a processor 6001, a parallel processing processor 6002, a bus 6003, a storage unit 6004, a RAM 6005, an interface unit 6006, and an interface unit 6007, which are connected to the bus 6003. The bus 6003 is, for example, a transmission path having a data transfer rate of several hundred MB / s or more. The bus 6003 may be a transmission path having a data transfer rate of 1 GB / s or more. The bus 3003 performs data transfer based on, for example, the PCI-Express or PCI-X standard. The analysis unit 600 may be connected to multiple measurement units 400 via multiple interface units 6006. When multiple measurement units 400 are provided, the analysis unit 600 may be connected to each of the measurement units 400 (for example, multiple measurement units 400 and multiple analysis units 600 are connected one-to-one).

[0260] As shown in FIG. 66, the processor 6001 and the parallel processing processor 6002 have the same configuration and functions as the processor 4831 and the parallel processing processor 4833 described above. The parallel processing processor 6002 includes multiple arithmetic units 6200 and a RAM 6201. Analysis software 6100 runs on the processor 6001 to analyze the cell type of measured cells. The analysis software 6100 running on the processor 6001 has the same functions as the analysis software 4835a shown in FIG. 52. The analysis software 6100 analyzes the cell type of measured cells by performing operations similar to those shown in FIGS. 52, 53(a) and 53(b), 54, 55(a) and 55(b), 56, 57, and related descriptions. The analysis software 6100 transmits classification information of the measured cells to the processing unit 300 via an interface unit 6007. The interface unit 6007 is, for example, Ethernet (registered trademark) or USB. The interface unit 6007 may be an interface capable of wireless communication (for example, WiFi (registered trademark), Bluetooth (registered trademark)).

[0261] FIG. 67 shows the configuration of the processing unit 300 in the fourth configuration example.

[0262] The processing unit 300 shown in FIG. 67 does not have to include a parallel processing processor 3002 like the processing unit 300 shown in FIGS. 59 and 62. Furthermore, the analysis software 3100 shown in FIGS. 59 and 62 does not have to run on the processor 3001 shown in FIG. 67. The processing unit 300 receives the analysis results from the analysis unit 600 via the interface unit 3006. The interface unit 3006 is, for example, Ethernet or USB. The interface unit 3006 may also be an interface capable of wireless communication (for example, WiFi or Bluetooth).

[0263] In the configurations of Figures 64, 65, 66, and 67, analog signals (forward scattered light signal, side scattered light signal, side fluorescent light signal) of cells generated in the FCM detection unit 410 are converted into digital signals in the A / D conversion unit 482 in the measurement unit 400. Waveform data is sent to the analysis unit 600 via the interface unit 484, bus 485, interface unit 4851, and transmission path 4852. As described above, the interface unit 4851 is a dedicated interface for connecting the measurement unit 400 and the analysis unit 600, and connects the measurement unit 400 and the analysis unit 600 one-to-one. In other words, the transmission path 4852 is a transmission path that does not involve the transmission of data related to devices other than the components constituting the cell analysis device 100 (e.g., the measurement unit 400 and the processing unit 300). The transmission path 4852 is a transmission path that is separate from, for example, an intranet or the Internet. This makes it possible to avoid bottlenecks in the communication speed of digital signal transmission even when digital signals generated by A / D conversion within the measurement unit 400 are transmitted to the processing unit 300.

[0264] In the case of the cell analyzer 100 of configuration example 4 shown in Figures 64, 65, 66, and 67, step S13 (cell classification) and step S14 (transmission of classification information) of the flowchart shown in Figure 10 are performed in the analysis unit 600. That is, the digital signal generated in step S13 is transmitted from the measurement unit 400 to the analysis unit 600, and in step S14 the classification information is transmitted to the processing unit 300. The processing in Figures 56 and 57 is performed by the processor 6001 and parallel processing processor 6002 of the analysis unit 600.

[0265] (Configuration Example 5) A fifth configuration example of the cell analyzer 100 will be described with reference to FIGS. 63, 67, and 68.

[0266] Like the aforementioned Configuration Example 4, the cell analyzer 100 of Configuration Example 5 is configured with a measuring unit 400, a processing unit 300, and an analysis unit 600. The measuring unit 400 of Configuration Example 5 shown in FIG. 68 has the same function and configuration as the measuring unit 400 described in FIG. 58 and its related description. The measuring unit 400 of Configuration Example 5 shown in FIG. 68 is connected to the analysis unit 600 via a connection cable 4202. For example, the connection cable 4202 is configured with a twisted pair cable and has a number of wire pairs corresponding to the type of analog signal transmitted to the processing unit 300. The connection cable 4202 is configured with a length of, for example, 1 meter or less to reduce noise during signal transmission. The measuring unit 400 transmits analog signals to the analysis unit 600 via the connection cable 4202.

[0267] The analysis unit 600 shown in Fig. 69 has the same functions and configuration as the analysis unit 600 described in Fig. 65 and its related description. That is, in the example of Fig. 69, a parallel processing processor 6002 is mounted on the cell analyzer 100 so as to be incorporated into the analysis unit 600. The analysis unit 600 shown in Fig. 69 further includes a connection port 6008 and an A / D conversion unit 6009. An analog signal transmitted from the analysis unit 600 via the connection cable 4202 is input to the A / D conversion unit 6009 via the connection port 6008. The A / D conversion unit 6009 converts the analog signal into a digital signal by processing similar to that of the A / D conversion unit 482.

[0268] The analysis unit 600 may be connected to a plurality of measurement units 400 via a plurality of connection ports 6008. When a plurality of measurement units 400 are provided, the analysis unit 600 may be connected to each of the measurement units 400 (for example, the plurality of measurement units 400 and the plurality of analysis units 600 are connected one-to-one).

[0269] As shown in FIG. 66, the processor 6001 and the parallel processing processor 6002 have the same configuration and functions as the processor 4831 and the parallel processing processor 4833 described above. Analysis software 6100 runs on the processor 6001, analyzing the cell type of the measured cells. The analysis software 6100 running on the processor 6001 has the same functions as the analysis software 4835a shown in FIG. 52. The analysis software 6100 analyzes the cell type of the measured cells by performing operations similar to those shown in FIGS. 52, 53(a) and 53(b), 54, 55(a) and 55(b), 56, 57, and their related descriptions. The analysis software 6100 transmits the analysis results of the cell type of the measured cells to the processing unit 300 via an interface unit 6007. The interface unit 6007 is, for example, Ethernet or USB. The interface unit 6007 may also be a wireless communication interface (for example, Wi-Fi or Bluetooth).

[0270] In the case of the cell analyzer 100 of Configuration Example 5 shown in Figures 68 and 69, the generation of waveform data and feature parameters in step S12 of the flowchart shown in Figure 10, step S13 (cell classification), and step S14 (transmission of classification information) are performed in the analysis unit 600. That is, the generation of waveform data (step S12) is performed in the A / D conversion section 6009 of the analysis unit 600, cell classification based on the digital signal (step S13) is performed by the processor 6001 and parallel processing processor 6002 of the analysis unit 600, and the classification information is transmitted from the analysis unit 600 to the processing unit 300 (step S14). The processing of Figures 56 and 57 is performed by the processor 6001 and parallel processing processor 6002 of the analysis unit 600.

[0271] 9. Neural Network Structure and Training for Deep Learning Algorithms 60 (Neural network structure) FIG. 70(a) illustrates the structure of a convolutional neural network that implements the deep learning algorithm 60.

[0272] The neural network includes an input layer 60a, an output layer 60b, and an intermediate layer 60c between the input layer 60a and the output layer 60b, and the intermediate layer 60c is composed of multiple layers. The number of layers constituting the intermediate layer 60c can be, for example, 5 or more, preferably 50 or more, and more preferably 100 or more.

[0273] In a neural network, multiple nodes 89 are arranged in layers and connected between layers, allowing information to propagate in only one direction, as indicated by arrow D in the figure, from the input layer 60a on the input side to the output layer 60b on the output side.

[0274] (operation at each node) FIG. 70(b) is a schematic diagram showing the calculations at each node.

[0275] Each node 89 receives multiple inputs and calculates one output (z). In the example shown in FIG. 70(b), node 89 receives four inputs. The total input (u) received by node 89 is expressed, for example, by the following (Equation 2). In this embodiment, one-dimensional row and column data is used as training input data and analysis input data, so if the variables of the arithmetic formula correspond to two-dimensional matrix data, processing is performed to convert the variables so that they correspond to one-dimensional matrix data.

[0276]

number

[0277] Each input is multiplied by a different weight. In (Equation 2), b is a value called the bias. The output (z) of the node is the output of a given function f for the total input (u) expressed in (Equation 2), and is expressed in the following (Equation 3). The function f is called the activation function.

[0278]

number

[0279] FIG. 70(c) is a schematic diagram showing operations between nodes.

[0280] In a neural network, nodes are arranged in layers, each of which outputs a result (z) expressed in (Equation 3) for the total input (u) of each node 89 expressed in (Equation 2). The output of a node in the previous layer becomes the input of a node in the next layer. In the example shown in Figure 70(c), the output of node 89a in the layer on the left side of the figure becomes the input of node 89b in the layer on the right side of the figure. Each node 89b receives the output from node 89a. A different weight is applied to each connection between each node 89a and each node 89b. If the outputs of each of the multiple nodes 89a are x1 to x4, the inputs to each of the three nodes 89b are expressed by the following (Equation 4-1) to (Equation 4-3).

[0281]

number

[0282] Generalizing these (Equation 4-1) to (Equation 4-3), we obtain the following (Equation 4-4), where i = 1, ..., I and j = 1, ..., J. I is the total number of inputs, and J is the total number of outputs.

[0283]

number

[0284] Applying (Equation 4-4) to the activation function gives the output, which is expressed as (Equation 5) below.

[0285]

number

[0286] (activation function) In the cell type analysis method according to the embodiment, a rectified linear unit function is used as the activation function. The rectified linear unit function is expressed by the following equation (6).

[0287]

number

[0288] (Equation 6) is a linear function of z=u, where the part of u<0 is set to u=0. In the example shown in FIG. 70(c), the output of the node j=1 is expressed by the following equation using (Equation 6):

[0289]

number

[0290] (Neural network training) If the function represented using a neural network is y(x:w), the function y(x:w) will change when the parameter w of the neural network is changed. Adjusting the function y(x:w) so that the neural network selects the parameter w that is more suitable for the input x is called neural network training. Suppose multiple pairs of input and output for the function represented using a neural network are given. If the desired output for a certain input x is d, the input / output pairs are given as {(x1, d1), (x2, d2), ..., (xn, dn)}. The set of pairs represented by (x, d) is called training data.

[0291] Learning a neural network means adjusting the weight w using an error function so that, for any input / output pair (xn, dn), when input xn is given, the neural network output y(xn:w) comes as close as possible to the output dn.

[0292]

number

[0293] An error function is a measure of the closeness between a function expressed using a neural network and training data. The error function is also called a loss function. The error function E(w) used in the cell type analysis method according to the embodiment is expressed by the following equation (7). Equation (7) is called cross entropy.

[0294]

number

[0295] A method for calculating the cross entropy of (Equation 7) will be explained. In the output layer of the neural network used in the cell type analysis method according to the embodiment, i.e., in the final layer of the neural network, an activation function is used to classify the input x into a finite number of classes according to its content. The activation function is called a softmax function, and is expressed by the following (Equation 8). It is assumed that the output layer 60b has the same number of nodes as the number of classes k. The total input u of each node k (k=1, ..., K) of the output layer L is calculated by dividing u by uk from the output of the previous layer L-1. (L) As a result, the output of the k-th node in the output layer is expressed as follows (Equation 8).

[0296]

number

[0297] (Equation 8) is the softmax function. The sum of the outputs y1, ..., yk determined by (Equation 8) is always 1.

[0298] If each class is represented as C1,...,Ck, the output yk of node k in the output layer L (i.e., uk (L) ) represents the probability that a given input x belongs to class Ck. Input x is classified into the class that maximizes the probability expressed by the following (Equation 9).

[0299]

number

[0300] In neural network training, the function represented by the neural network is regarded as a model of the posterior probability of each class, and under such a probability model, the likelihood of the weight w for the training data is evaluated, and the weight w that maximizes the likelihood is selected.

[0301] The target output dn by the softmax function in (Equation 8) is set to 1 only if the output is the correct class, and to 0 otherwise. If the target output is expressed in vector form as dn = [dn1, ..., dnk], for example, if the correct class of the input xn is C3, only the target output dn3 will be 1, and the other target outputs will be 0. When encoded in this way, the posterior distribution is expressed by the following (Equation 10).

[0302]

number

[0303] The likelihood L(w) of weight w for training data {(xn, dn)} (n = 1, ..., N) is expressed as follows (Equation 11): Taking the logarithm of the likelihood L(w) and inverting the sign yields the error function of (Equation 7).

[0304]

number

[0305] Learning refers to minimizing the error function E(w) calculated based on training data with respect to the parameter w of the neural network. In the cell type analysis method according to the embodiment, the error function E(w) is expressed by Equation 7.

[0306] Minimizing the error function E(w) with respect to the parameter w is equivalent to finding a local minimum of the function E(w). The parameter w is the weight of the connection between nodes. The minimum of the weight w is found by iterative calculations that start with an arbitrary initial value and repeatedly update the parameter w. An example of such calculations is the gradient descent method.

[0307] The gradient descent method uses a vector expressed by the following (Equation 12).

[0308]

number

[0309] Gradient descent involves repeatedly moving the current value of the parameter w in the negative gradient direction (i.e., -∇E). (t) The weight after the movement is w (t+1) Then, the calculation by the gradient descent method is expressed by the following (Equation 13): The value t indicates the number of times the parameter w is moved.

[0310]

number

[0311] The symbol shown in the following (Equation 14) used in the above (Equation 13) is a constant that determines the magnitude of the update amount of the parameter w, and is called a learning coefficient.

[0312]

number

[0313] By repeating the calculation expressed by (Equation 13), the error function E(w (t) ) decreases and the parameter w reaches a minimum.

[0314] The calculation using (Equation 13) may be performed on all training data (n=1, ..., N) or only on a portion of the training data. A gradient descent method performed on only a portion of the training data is called a stochastic gradient descent method. The cell type analysis method according to the embodiment uses the stochastic gradient descent method.

[0315] An overview of training of the deep learning algorithm 60 will be explained using Figure 71 as an example. As described above, the deep learning algorithm 60 is composed of a neural network including multiple hidden layers. As training data, waveform data of cells whose cell types have been previously identified is input to the input layer 60a of the neural network. As training data, probability data 78 is input to the output layer 60b of the neural network. If the number of cell types to be classified by the deep learning algorithm 60 is nine as shown in Figure 8, the output layer 60b will have nine nodes, and each node will be assigned a cell type. The probability data 78 is a data group in which the probability of the label value of the previously identified cell type corresponding to the waveform data input to the input layer 60a is 100% and the probability of other cell types is 0%. In this way, training data is input to the input layer 60a and the output layer 60b, respectively, and the neural network is trained.

[0316] [10. Effects of the embodiment] According to the above-described embodiment, test result data is generated that includes the results of classifying individual cells into multiple cell types, as representatively illustrated in, for example, Figure 12. In conventional cell classification methods, cells with characteristics of, for example, neutrophils and immature granulocytes are classified alternatively into one of these types. However, in the above-described embodiment, the probability that such cells belong to either neutrophils or immature granulocytes is calculated using a deep learning algorithm, and the probability values ​​are stored in the test result data. Therefore, cells that were previously classified alternatively into one type can now be classified into multiple cell types. [Explanation of symbols]

[0317] 60 Deep Learning Algorithms 86a, 86b, 86c Waveform data 100 Cell analyzer (Analyzer) 410 FCM detector (detector) 482 A / D conversion unit (signal processing unit) 901 Scattergram (graph) 3001, 4831 processors 3002, 4833 parallel processing processor 4113 Flow Cell

Claims

1. An analytical method for analyzing a sample containing cells, comprising: Irradiating a measurement sample prepared from the specimen with light and detecting light emitted from the cells; acquiring characteristic data of each of a plurality of cells contained in the specimen based on the detected light; By analyzing the feature data using a deep learning algorithm, a probability that each of the cells corresponds to each of a plurality of cell types is determined, and a first cell type and a second cell type are determined based on the probability; generating result data including (1) a first analysis result obtained by counting and / or classifying the plurality of cells based on the first cell type, and (2) a second analysis result based on the second cell type, in association with the ID of the specimen; An analytical method characterized by:

2. the plurality of cell types includes at least one normal cell and at least one abnormal cell; The analytical method according to claim 1 , wherein the result data includes a probability that each of the cells belongs to the normal cell and the abnormal cell.

3. The analysis method according to claim 1 or 2, wherein the deep learning algorithm is trained using cell feature data and classifications corresponding to the cells as training data.

4. The analysis method according to any one of claims 1 to 3, wherein the plurality of cell types include lymphocytes, monocytes, eosinophils, neutrophils, basophils and abnormal blood cells.

5. The analysis method according to claim 4 , wherein the abnormal blood cells include at least one of immature granulocytes, blast cells, and abnormal lymphocytes.

6. The analysis method according to any one of claims 1 to 5, wherein the result data further includes identification information assigned to the first cell type.

7. The analysis method according to any one of claims 1 to 6, wherein the result data further includes a flag for distinguishing the first cell type from other cell types.

8. A ranking is determined for the plurality of cell types in descending order of the probability that a single cell corresponds to the plurality of cell types; The analysis method according to any one of claims 1 to 7, wherein the result data includes the ranking.

9. the detecting includes detecting light generated by the cells passing through an illuminated flow cell; 9. The analysis method according to claim 1, wherein the acquisition of the characteristic data includes acquiring a waveform signal that changes over time in response to the detected light.

10. The analysis method according to claim 9 , wherein the feature data is waveform data generated by sampling the waveform signal at multiple points in time.

11. The analysis method according to any one of claims 1 to 10, wherein the analysis of the feature data using the deep learning algorithm is performed using a processor and a parallel processing processor operating under the instructions of the processor.

12. The analysis method according to any one of claims 1 to 11, further comprising obtaining characteristic parameters for plotting cells on a graph based on the characteristic data.

13. An analytical device for analyzing a sample containing cells, a detection unit that irradiates a measurement sample prepared from a specimen with light and detects light emitted from the cells; a signal processing unit that acquires characteristic data of each of a plurality of cells contained in the specimen based on the detected light; an information processing unit including at least one processor; The processor: By analyzing the feature data using a deep learning algorithm, a probability that each of the cells corresponds to each of a plurality of cell types is determined, and a first cell type and a second cell type are determined based on the probability; generating result data including (1) a first analysis result obtained by counting and / or classifying the plurality of cells based on the first cell type, and (2) a second analysis result based on the second cell type, in association with the ID of the specimen; An analytical device characterized by:

14. the plurality of cell types includes at least one normal cell and at least one abnormal cell; The analytical device of claim 13 , wherein the result data includes a probability that the cell belongs to the normal cell and the abnormal cell.

15. The analysis device according to claim 13 or 14, wherein the deep learning algorithm is trained using cell feature data and classifications corresponding to the cells as training data.

16. The analyzer according to any one of claims 13 to 15, wherein the plurality of cell types include lymphocytes, monocytes, eosinophils, neutrophils, basophils, and abnormal blood cells.

17. The analyzer of claim 16, wherein the abnormal blood cells include at least one of immature granulocytes, blast cells, and abnormal lymphocytes.

18. The analytical device according to any one of claims 13 to 17, wherein the result data generated by the processor further includes identification information assigned to the first cell type.

19. The analyzer according to any one of claims 13 to 18, wherein the result data generated by the processor further includes a flag for distinguishing the first cell type from other cell types.

20. the processor determines a ranking for the plurality of cell types in descending order of the probability that the cell corresponds to the plurality of cell types; The analysis device according to any one of claims 13 to 19, wherein the result data generated by the processor includes the ranking.

21. the detection unit detects light generated when cells pass through a flow cell irradiated with light; The analysis device according to any one of claims 13 to 20, wherein the signal processing unit acquires, as the feature data, a waveform signal that changes over time in accordance with the detected light.

22. The analysis device according to claim 21 , wherein the feature data is waveform data generated by sampling the waveform signal at multiple points in time.

23. The analysis device according to any one of claims 13 to 22, wherein the processor uses a parallel processing processor operating under the instruction of the processor to analyze the feature data using the deep learning algorithm.

24. The analyzer according to any one of claims 13 to 23, wherein the signal processing unit acquires characteristic parameters for plotting cells on a graph based on the characteristic data.

Images