Disease differentiation support method, disease differentiation support device, and disease differentiation support computer program

Abstract

To provide a method, a device, and a computer program for supporting differentiation of a disease that allow support of differentiation of a disease by a simple pre-treatment of a sample.SOLUTION: The present invention relates to a method for supporting differentiation of a disease which includes: acquiring a first parameter group including a first parameter obtained by analyzing an image including cells in a sample taken from a subject; acquiring a second parameter group including a second parameter obtained by analyzing an optical signal or an electric signal obtained from cells in the sample; and generating differentiation support information for supporting differentiation of a disease on the basis of the first parameter group and the second parameter group by using a computer algorithm.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] This specification discloses a disease differentiation support method, a disease differentiation support device, and a disease differentiation support computer program. [Background technology]

[0002] Disease differentiation is performed by testing samples obtained from subjects. For example, Non-Patent Document 1 discloses a method for differentiating myeloproliferative neoplasms (MPN) from polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF) based on gene expression obtained by next-generation sequencing. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Manja Meggendorfer et al., Deep Learning Algorithms Support Distinction of PV, PMF, and ET Based on Clinical and Genetic Markers, Blood 2017 130:4223 Summary of the Invention [Problem to be solved by the invention]

[0004] However, conventional disease differentiation tests require complicated testing procedures and require the expertise of experienced technicians, creating a need for new methods to support disease differentiation.

[0005] An object of the present invention is to provide a new disease differentiation support method, a disease differentiation support device, and a disease differentiation support computer program that enable support in disease differentiation. [Means for solving the problem]

[0006] This specification discloses a disease differentiation support method for supporting disease differentiation, which includes acquiring a first parameter obtained by analyzing an image including cells contained in a sample collected from a subject, acquiring a second parameter related to the number of cells contained in the sample, and generating differentiation support information for supporting disease differentiation based on the first parameter and the second parameter using a computer algorithm.

[0007] This specification discloses a disease differentiation support method for supporting disease differentiation, which uses a computer algorithm to generate differentiation support information for supporting disease differentiation based on a first parameter obtained by analyzing an image including cells contained in a sample collected from a subject and a second parameter related to the number of cells contained in the sample.

[0008] This specification discloses a disease differentiation support device (200A, 200B, 100B) for supporting disease differentiation. The disease differentiation support device (200A, 200B, 100B) includes a processing unit (20A, 20B, 10B). The processing unit (20A, 20B, 10B) acquires a first parameter obtained by analyzing an image including cells contained in a sample collected from a subject, acquires a second parameter related to the number of cells contained in the sample, and generates differentiation support information for supporting disease differentiation based on the first parameter and the second parameter using a computer algorithm.

[0009] This specification discloses a computer program for assisting in the differentiation of diseases, which, when executed by a computer, executes a process including the steps of acquiring a first parameter obtained by analyzing an image including cells contained in a sample collected from a subject, acquiring a second parameter related to the number of cells contained in the sample, and generating differentiation assistance information for assisting in the differentiation of diseases based on the first parameter and the second parameter using a computer algorithm.

[0010] The disease differentiation support method, differentiation support device, and computer program disclosed in this specification can support disease differentiation by utilizing information obtained by testing methods routinely performed in clinical laboratories, etc. Effect of the Invention

[0011] A new disease differentiation support method, a disease differentiation support device, and a disease differentiation support computer program are provided. [Brief description of the drawings]

[0012] [Figure 1] Methods to aid in disease differentiation are outlined. [Diagram 2] An example of the first parameter group is shown below. [Diagram 3] An example of the second parameter group is shown below. [Figure 4] An example of training data is shown below. [Diagram 5] 1 shows an example of the configuration of a disease differentiation support system 1. [Figure 6] 1 shows an outline of the configuration of a cell image analyzer. [Figure 7] 1 shows an example of the configuration of a blood cell counter equipped with an optical detection unit. [Figure 8] 1 shows an example of the configuration of a blood cell counter equipped with an electrical resistance type detection unit. [Figure 9] 1 shows the hardware configuration of a training device 100A and a disease differentiation support device 100B. [Figure 10] 2 shows an example of a functional configuration of a training device 100A. [Figure 11] 4 shows the processing flow of the training program. [Figure 12] 2 shows the hardware configuration of disease differentiation support devices 200A and 200B and a terminal device 200C. [Figure 13] 2 shows an example of a functional configuration of a disease differentiation support device 200A. [Figure 14] 1 shows the processing flow of a computer program for assisting in disease differentiation. [Figure 15] 2 shows an example of the configuration of a disease differentiation support system 2. [Figure 16] 1 shows an example of a functional configuration of a disease differentiation support device 200B. [Figure 17] 1 shows an example of the configuration of a disease differentiation support system 3. [Figure 18] 1 shows an example of a functional configuration of a disease differentiation support device 100B. [Figure 19] A comparison of disease prediction results by machine methods and doctor's diagnoses is shown. [Figure 20] The ROC curve of the prediction results of PV by the machine method is shown. [Figure 21] The ROC curve of the machine-based ET prediction results is shown. [Figure 22] The ROC curve of the PMF prediction results by the machine method is shown. [Figure 23] The prediction accuracy of the mechanical method is shown. [Figure 24] The prediction results by the algorithm described in Non-Patent Document 1 are shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] 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 indicate the same or similar components, and therefore, the description of the same or similar components will be omitted.

[0014] 1. Methods to aid in disease differentiation 1-1. Overview of support methods This embodiment relates to a disease differentiation support method (hereinafter, simply referred to as "support method") for supporting disease differentiation. The support method acquires a plurality of first parameters and a plurality of second parameters related to cells from a sample containing cells collected from a subject, and generates differentiation support information for supporting disease differentiation based on the plurality of first parameters and the plurality of second parameters using a computer algorithm.

[0015] The first parameter is obtained from an analysis result obtained by analyzing an image including cells contained in a sample collected from a subject. The second parameter is a parameter related to the number of cells, and is obtained from an analysis result obtained by analyzing an optical signal or an electrical signal obtained from the cells contained in the sample. In this specification, the parameter related to the number of cells includes, in addition to the number of cells, the concentration of a specific cell per a given amount of sample (e.g., the concentration of red blood cells per 1 μL), which is a value calculated based on the number of cells, and the ratio of a specific cell to a certain cell (e.g., the ratio of eosinophils per 100 white blood cells). FIG. 1 shows an outline of the support method. As shown in FIG. 1, a sample S collected from a subject is divided into a specimen S1 to be used for analysis in step A and a specimen S2 to be used for analysis in step B. In step A, a cell image analyzer 400 captures an image of cells contained in the specimen S1, and the obtained cell image P is analyzed to obtain a first parameter group consisting of a plurality of first parameters including a parameter related to an abnormal finding. In step B, a blood cell counter 450 obtains an optical signal S or an electrical signal from cells contained in the specimen S2, and the obtained optical signal L or electrical signal is analyzed to obtain a second parameter group consisting of a plurality of second parameters including the number or ratio of each type of cell. In step C, the first parameter group consisting of a plurality of first parameters and the second parameter group consisting of a plurality of second parameters are input to a pre-trained computer algorithm CA to generate discrimination support information for supporting the discrimination of diseases.

[0016] In this embodiment, the cells contained in the sample belong to a predetermined cell group. The predetermined cell group is a cell group that constitutes each human organ. The predetermined cell group normally includes a plurality of cell types that are morphologically classified by histological or cytological microscopic observation. Morphological classification (also called "morphological classification") includes classification of cell types and classification of morphological characteristics of cells. Preferably, the cells to be analyzed are a cell group that belongs to a predetermined cell lineage that belongs to the predetermined cell group. The predetermined cell lineage is a cell group that belongs to the same lineage differentiated from a certain type of tissue stem cell. The predetermined cell lineage is preferably the hematopoietic system, and the hematopoietic cells are more preferably peripheral blood cells or bone marrow cells.

[0017] Generally, in a blood test, a hematopoietic sample S taken from a subject is used to measure the number of red blood cells, white blood cells, platelet count, hemoglobin concentration, hematocrit value, red blood cell constants, white blood cell differential value, etc. In addition to the blood cell test using the hematopoietic sample counter, if a blood disease is suspected, a smear is prepared from the sample S and the actual morphological observation of the blood cells is performed to check for morphological abnormalities in the cells.

[0018] (1) First parameter group In this embodiment, in step A shown in FIG. 1, in order to obtain the first parameter, morphological features are extracted from the cells in an image captured under a microscope or by a slide scanner for each cell on a smear prepared from the specimen S1.

[0019] For the extraction of morphological characteristics, it is preferable to use a specimen that has been subjected to bright-field staining. The bright-field staining is preferably selected from Wright's staining, Giemsa staining, Wright-Giemsa staining, and May-Giemsa staining. May-Giemsa staining is more preferable. The specimen is not limited as long as it allows the morphology of each cell belonging to a predetermined cell group to be observed individually. Examples include smear specimens and imprint specimens. A smear specimen made from peripheral blood or bone marrow is preferable.

[0020] In the extraction of morphological features from cells, individual cells on a specimen are morphologically classified. If the cells have abnormal findings, the abnormal findings are classified. By the morphological classification of the cells, at least one of the type of cell and the ratio of cells of the same type contained in the sample is obtained as a parameter related to the morphological classification of the cells. By the classification of the abnormal findings, at least one of the type of abnormal findings and the ratio of cells exhibiting the same type of abnormal findings is obtained as a parameter related to the abnormal findings. The first parameter is at least one of a parameter related to the morphological classification and a parameter related to the abnormal findings. The first parameter group is preferably a group of first parameters detected in one observation. An example of the first parameter group is shown in FIG. 2.

[0021] Parameters relating to abnormal findings can include, for example, values ​​related to at least one selected from nuclear morphological abnormalities, granular abnormalities, abnormal cell size, cell malformations, cell destruction, vacuoles, immature cells, presence of inclusion bodies, Döhle bodies, satellite phenomenon, nuclear reticulum abnormalities, petal-like nuclei, large N / C ratio, bleb-like morphology, smudges, and hairy cell-like morphology.

[0022] Nuclear morphological abnormalities may include at least one selected from hypersegmentation, hyposegmentation, pseudo-Pelger anomaly, ring-shaped nucleus, spherical nucleus, oval nucleus, apoptosis, multinuclei, karyorrhexis, enucleation, bare nucleus, irregular nuclear margins, nuclear fragmentation, internuclear bridges, multiple nuclei, cleaved nuclei, nuclear division, and nucleolar abnormalities.

[0023] The granular abnormality may include at least one selected from degranulation, granule distribution abnormality, toxic granules, Auer rods, Bassoon cells, and pseudo-Chediak-Higashi granule-like granules. Cell size abnormalities can include giant platelets.

[0024] The parameters relating to the morphological classification of cells may include a value related to the number of at least one type of cell selected from neutrophils, eosinophils, platelets, lymphocytes, monocytes, basophils, metamyelocytes, myelocytes, promyelocytes, blasts, plasma cells, atypical lymphocytes, immature eosinophils, immature basophils, erythroblasts, and megakaryocytes, and a value related to at least one of the ratios of at least one type of cell selected from neutrophils, eosinophils, platelets, lymphocytes, monocytes, basophils, metamyelocytes, myelocytes, promyelocytes, blasts, plasma cells, atypical lymphocytes, immature eosinophils, immature basophils, erythroblasts, and megakaryocytes.

[0025] The method of acquiring the first parameter is not limited as long as the above-mentioned parameters related to morphological classification and / or parameters related to abnormal findings can be acquired. For example, the first parameter may be acquired by an examiner, but it is preferable to use the cell image analyzer 300 or 400 described later.

[0026] The first parameter may be acquired using a deep learning algorithm described in US Patent Publication No. 2019-0347467 and the like. US Patent Publication No. 2019-0347467 is incorporated herein by reference. The classifier used in the method for acquiring the first parameter includes a plurality of deep learning algorithms having a neural network structure, as shown in FIG. 1. The classifier includes a first deep learning algorithm DL1 and a second deep learning algorithm DL2, and the first deep learning algorithm DL1 extracts a feature amount of a cell, and the second deep learning algorithm DL2 identifies the cell to be analyzed based on the feature amount extracted by the first deep learning algorithm. The second deep learning algorithm DL2 outputs a morphological classification result and a probability of falling into the classification, or a classification of abnormal findings and a probability of falling into the classification, for each cell. The first deep learning algorithm DL1 is a convolution-connected neural network, and the second deep learning algorithm DL2, which is downstream of the first deep learning algorithm, is a full-connected neural network.

[0027] (2) Second parameter In this embodiment, in step B shown in FIG. 1, the sample S2 is measured using a blood cell counter to obtain the second parameter.

[0028] The second parameter is a parameter obtained based on an optical signal or an electrical signal detected by the hemocytometer 350, 450, and is a value related to at least one of the number of each type of cell, the ratio of each type of cell, the size of the cell, and the concentration of a component contained in the cell. The second parameter may be obtained using a deep learning algorithm described in U.S. Patent Publication No. 2014-0051071, etc. U.S. Patent Publication No. 2014-0051071 is incorporated herein by reference. Examples of parameters obtained from the blood cell counters 350, 450 are shown in FIG.

[0029] The second parameter may preferably include (i) a value related to the number of at least one type of cell selected from red blood cells, nucleated red blood cells, small red blood cells, platelets, hemoglobin, reticulocytes, immature granulocytes, neutrophils, eosinophils, basophils, lymphocytes, and monocytes, (ii) a value related to the ratio of at least one type of cell selected from red blood cells, nucleated red blood cells, small red blood cells, platelets, hemoglobin, reticulocytes, immature granulocytes, neutrophils, eosinophils, basophils, lymphocytes, and monocytes, and (iii) a value related to at least one selected from hematocrit, mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), and mean platelet volume (MPV). The second parameter may include at least one parameter selected from (i), (ii), and (iii).

[0030] (3) Differentiation of diseases In the present embodiment, differentiation support information for supporting the differentiation of diseases is generated based on a first parameter group and the second parameter group.

[0031] The disease to be differentiated in this embodiment is not limited as long as it is a human disease. Preferably, it is a hematopoietic disease. Hematopoietic diseases may include myeloproliferative neoplasms, leukemia, myelodysplastic syndromes, lymphomas, myelomas, and the like. Myeloproliferative neoplasms may preferably include polycythemia vera, essential thrombocythemia, or primary myelofibrosis, and the like. Leukemias may preferably include acute myeloblastic leukemia, acute myeloblastic leukemia, acute promyelocytic leukemia, acute myelomonocytic leukemia, acute monocytic leukemia, erythroleukemia, acute megakaryoblastic leukemia, acute myelogenous leukemia, acute lymphocytic leukemia, lymphoblastic leukemia, chronic myelogenous leukemia, and chronic lymphocytic leukemia, and the like. Lymphomas may include Hodgkin's lymphoma and non-Hodgkin's lymphoma, and the like. Myelomas may include multiple myeloma, and the like.

[0032] The support information for distinguishing diseases is generated by the processing unit 20, which will be described later, using a computer algorithm. The computer algorithm may include a machine learning algorithm, a deep learning algorithm, and the like.

[0033] The machine learning algorithms may include algorithms such as trees, regression, support vector machines, Bayesian, clustering, and random forests. A preferred machine learning algorithm is the gradient boosting tree algorithm. The gradient boosting tree is more preferably a multimodal deep neural network (MultimodalDNN). The deep learning algorithm has a neural network structure. A computer algorithm is trained according to the method described below to function as a disease discriminator.

[0034] 1-2. Generation of classifier Training data is generated by arranging the first training parameter group and the second training parameter group in a matrix of the same hierarchy, and further linking them with a label indicating the disease name (hereinafter also referred to as "disease name label"). The first training parameter group is obtained from a training smear. An example of training data is shown in Figure 4. For example, the first column in Figure 4 indicates the category of the parameter group. The second column in Figure 4 indicates the row number. The third column in Figure 4 indicates the name (label) of each parameter, and the fourth column indicates each parameter. Parameters may be expressed, for example, in %, for the ratio of cells with cell types or abnormal findings, in g / dL, for the concentration, and in 10^4 / uL (×10 4 The first parameter is expressed in units such as 1 / μL. These units are commonly used in blood cell tests and the like. The first parameter is obtained by using the deep learning algorithm DL2 to count the results of morphological classification or abnormal findings classification, and weighting the obtained counting result by the probability that it corresponds to the classification. For example, if the cell classified as a neutrophil band cell is 1 out of 100 cells, and the probability that the cell classified as a neutrophil band cell is classified as a neutrophil band cell is 90%, multiply 1 by 90% to get 0.9. The fifth column in Figure 4 is a label indicating the disease name.

[0035] The first training parameter group and the second training parameter group are generated for each sample (hereinafter also referred to as "training sample") taken from a patient whose disease has been definitively diagnosed by a doctor. Then, for each sample, a matrix consisting of the first training parameter group and the second training parameter group is associated with the disease name to become training data.

[0036] In the example of Fig. 4, the first training parameter group and the second training parameter group are arranged in the row direction, but they may be arranged in the column direction. Also, each parameter may be represented by an abbreviation as in Fig. 4, or may be represented by a label value. Also, the label of the disease name may be represented by a label value.

[0037] Here, the first training parameter group and the second training parameter group may be subjected to a predetermined statistical analysis to select parameters highly related to the disease. Examples of the predetermined statistical analysis include one-dimensional analysis of variance (ANOVA), Pearson's correlation, and Spearman's rank correlation. One-dimensional analysis of variance is preferable. By selecting statistical parameters, the accuracy of the discrimination can be improved.

[0038] The training data is then input to a computer algorithm to train the computer algorithm and generate a classifier. When using a machine learning algorithm, one algorithm is trained for each disease. On the other hand, when using a deep learning algorithm, one algorithm can be trained for multiple diseases.

[0039] Training of the computer algorithm can be done using software such as Python.

[0040] Trained computer algorithms are used as disease discriminators to aid in disease differentiation.

[0041] 1-3. Generation of analytical data and identification support information The analysis data is generated by acquiring a first group of analysis parameters and a second group of analysis parameters from an analysis target sample collected from a subject, and combining them. The first group of analysis parameters and the second group of analysis parameters are preferably generated in the same manner as the first group of training parameters and the second group of training parameters, respectively. In addition, the parameters included in the analysis data are preferably the same type as the parameters included in the training data. The analysis data is preferably generated by forming the first group of analysis parameters and the second group of analysis parameters into a matrix in the same hierarchy, preferably in the same order as the training data.

[0042] Next, the analysis data is input to the classifier trained in 1-2 above to generate information for assisting in disease discrimination. The information is a value indicating the probability that the subject has the disease corresponding to the classifier, predicted from the analysis data. Furthermore, a label indicating the name of the patient's disease may be output based on the probability.

[0043] 2.Disease Differentiation Support System 1 An embodiment of the present disclosure relates to a disease differentiation support system 1. 5, the configuration of the disease differentiation support system 1 will be described. The disease differentiation support system 1 includes a training device 100A and a disease differentiation support device 200A. The vendor-side device 100 functions as the training device 100A, and the user-side device 200 operates as the disease differentiation support device 200A.

[0044] The training device 100A is connected to a cell image analyzer 300 and a blood cell counter 350. The training device 100A acquires a first group of training parameters from the cell image analyzer 300, and acquires a second group of training parameters from the blood cell counter 350.

[0045] The disease differentiation support device 200A is connected to a cell image analyzer 400 and a blood cell counter 450. The disease differentiation support device 200A acquires a first group of analysis parameters from the cell image analyzer 400 and acquires a second group of analysis parameters from the blood cell counter 350.

[0046] The recording medium 98 is a computer-readable non-volatile recording medium, such as a DVD-ROM or a USB memory. Each component will be described below.

[0047] 2-1. Cell image analysis device The configuration of the cell-image analyzing device 300 will be described with reference to FIG. 6. The cell-image analyzing device 300 includes at least an imaging section 304, which includes a stage 309 for placing a specimen, a magnifying glass section 302 such as a microscope, and an imaging element 301 for capturing a microscopic image. An image of each cell on a training specimen 308 set on the stage 309 is acquired. The cell-image analyzing device 300 acquires a first parameter group from the acquired image. The cell-image analyzing device 300 includes an information processing unit 305, which acquires and writes the first parameter group and communicates with the training device 100A.

[0048] Next, the configuration of the cell image analyzer 400 will be described. The configuration of the cell image analyzer 400 is basically the same as that of the cell image analyzer 300, and includes an imaging unit 404, which includes a stage 409 for placing a specimen, a magnifying glass unit 402 such as a microscope, and an imaging element 401 for capturing a microscopic image. The cell image analyzer 400 acquires an image of each cell on an analysis specimen 408 set on the stage 409. The cell image analyzer 400 acquires a first group of analysis parameters from the acquired image. The cell image analyzer 400 includes an information processing unit 405, which acquires and writes the first group of parameters and communicates with the disease differentiation support device 200.

[0049] As the cell image analyzers 300 and 400, for example, an Automated Digital Cell Morphology Analyzer DI-60 manufactured by Sysmex Corporation can be used.

[0050] 2-2. Hematology analyzer 7 and 8, the configuration of the blood cell counting device 350 will be described. The blood cell counting device 350 is a flow cytometer or the like that is provided with an optical detection unit 411 for detecting an optical signal and that is provided with a flow cell shown in FIG. In FIG. 7, light emitted from a laser diode, which is a light source 4111 , is irradiated via an illumination lens system 4112 onto cells passing through a flow cell 4113 .

[0051] In this embodiment, the light source 4111 of the flow cytometer is not particularly limited, and a light source 4111 having a wavelength suitable for exciting a fluorescent dye is selected. As such a light source 4111, for example, a semiconductor laser including a red semiconductor laser and / or a blue semiconductor laser, a gas laser such as an argon laser or a helium-neon laser, a mercury arc lamp, etc. are used. In particular, a semiconductor laser is preferable because it is very inexpensive compared to a gas laser.

[0052] As shown in FIG. 7, forward scattered light emitted from a particle passing through a flow cell 4113 is received by a forward scattered light receiving element 4116 via a condenser lens 4114 and a pinhole portion 4115. The forward scattered light receiving element 4116 may be a photodiode or the like. The side scattered light is received by a side scattered light receiving element 4121 via a condenser lens 4117, a dichroic mirror 4118, a bandpass filter 4119, and a pinhole portion 4120. The side scattered light receiving element 4121 may be a photodiode, a photomultiplier, or the like. The side fluorescent light is received by a side fluorescent light receiving element 4122 via the condenser lens 4117 and a dichroic mirror 4118. The side fluorescent light receiving element 4122 may be an avalanche photodiode, a photomultiplier, or the like.

[0053] The received light signals output from the light receiving parts 4116 , 4121 , and 4122 are subjected to analog processing such as amplification and waveform processing by analog processing parts having amplifiers 4151 , 4152 , and 4153 , respectively, and are sent to the measurement unit control part 480 .

[0054] The measurement unit control section 480 is connected to the information processing unit 351 , and the information processing unit 351 acquires a second parameter based on the optical signal acquired in the optical detection section 411 .

[0055] The blood cell counter 350 may also include an electric resistance type detection unit 412 as shown in FIG. 8. FIG. 8 shows a case where the electric resistance type detection unit 412 is a sheath flow type electric resistance detection unit 412. The sheath flow type electric resistance detection unit 412 includes a chamber wall 412a, an aperture unit 412b for measuring the electric resistance of cells, a sample nozzle 412c for supplying a sample, and a recovery tube 412d for recovering cells that have passed through the aperture unit 412b. The sample nozzle 412c and the recovery tube 412d in the chamber wall 412a are filled with sheath fluid. The dashed arrow indicated by the symbol 412s indicates the direction in which the sheath fluid flows. Red blood cells 412e and platelets 412f discharged from the sample nozzle pass through the aperture unit 412b while being enveloped in the flow of sheath fluid 412s. A constant DC voltage is applied to the aperture unit 412b, and it is controlled so that a constant current flows while only the sheath fluid flows. Since cells do not easily conduct electricity, i.e., have a large electrical resistance, when a cell passes through aperture portion 412b, the electrical resistance changes, and the number of times the cell has passed through aperture portion 412b and the electrical resistance can be detected. Since electrical resistance increases in proportion to the volume of the cell, measurement unit information processing portion 481 can calculate the volume of the cell that has passed through aperture portion 412b from the signal intensity related to the electrical resistance value, and send the count number of cells for each volume to information processing unit 351 as a histogram. The blood cell counter 350 measures the training sample and obtains a second set of training parameters.

[0056] The configuration of the blood cell counter 450 is similar to that of the blood cell counter 350. The blood cell counter 450 measures a training sample and obtains a second group of analysis parameters. An example of the blood cell counters 350 and 450 is the XN-2000 blood cell counter manufactured by Sysmex Corporation.

[0057] 2-3.Training device The training device 100A trains a computer algorithm using a first group of training parameters, a second group of training parameters, and disease names associated with these as training data to generate a classifier. The training device 100A acquires the first parameters from the cell image analyzer 300 through a recording medium 98 or a network 99. The training device 100A acquires the second parameters from the blood cell counter 350 through a recording medium 98 or a network 99. The training device 100A provides the generated classifier to the disease discrimination support device 200A. The classifier is provided through the recording medium 98 or the network 99. The disease discrimination support device 200A uses the classifier to generate discrimination support information for supporting disease discrimination.

[0058] (1) Hardware configuration of the training device The hardware configuration of the training device 100A will be described with reference to Figure 9. The training device 100A includes a processing unit 10 (10A), an input unit 16, and an output unit 17. The training device 100A is configured, for example, by a general-purpose computer.

[0059] The processing unit 10 includes a CPU (Central Processing Unit) 11 that performs data processing described later, a memory 12 used as a working area for data processing, a storage unit 13 that records a program and processing data described later, a bus 14 that transmits data between each unit, an interface unit 15 that inputs and outputs data to and from an external device, and a GPU (Graphics Processing Unit) 19. An input unit 16 and an output unit 17 are connected to the processing unit 10. Illustratively, the input unit 16 is an input device such as a keyboard, a touch panel, or a mouse, and the output unit 17 is a display device such as a liquid crystal display. The GPU 19 functions as an accelerator that assists the arithmetic processing (for example, parallel arithmetic processing) performed by the CPU 11. That is, in the following description, the processing performed by the CPU 11 includes the processing performed by the CPU 11 using the GPU 19 as an accelerator.

[0060] The processing unit 10 also stores in advance in the storage unit 13 a computer program for performing the training process described below in Fig. 11 and a computer algorithm, for example in an executable format. The executable format is a format generated by converting a programming language using a compiler, for example. The processing unit 10 cooperates with an operation system stored in the storage unit 13 to perform the training process of the computer algorithm using a computer program for performing the training process (hereinafter, sometimes simply referred to as a "training program").

[0061] In the following description, unless otherwise specified, the processing performed by the processing unit 10 means processing performed by the CPU 11 based on the computer program and computer algorithm for performing training processing stored in the storage unit 13 or memory 12. The CPU 11 uses the memory 12 as a working area to temporarily store necessary data (intermediate data during processing, etc.) in a volatile manner, and appropriately records data to be stored for a long time, such as calculation results, in the storage unit 13 in a non-volatile manner.

[0062] (2) Functional configuration of the training device 10, the processing unit 10A of the training device 100A functions as a training data generating unit 101, a training data input unit 102, and an algorithm updating unit 103. These functions are realized by installing a training program (e.g., Python, etc.) that causes a computer to execute a training process in the storage unit 13 or memory 12 of the processing unit 10A and executing this program by the CPU 11. The training data database (DB) 104 stores a first group of training parameters acquired by the processing unit 10A from the cell image analyzer 300 and a second group of training parameters acquired from the blood cell counter 350. The training data DB 104 also stores labels of disease names corresponding to the parameters. The algorithm database (DB) 105 can store computer algorithms before training and computer algorithms after training.

[0063] The training data generating unit 101 corresponds to step S11, the training data input unit 102 corresponds to step S12, and the algorithm updating unit 103 corresponds to step S15, which will be described later.

[0064] (3) Processing of training programs The processing unit 10A of the training device 100A executes each step of the training program shown in FIG.

[0065] The processing unit 10A receives a training data acquisition process start command input by the operator from the input unit 16, and in step S11, acquires a first training parameter group from the cell image analyzer 300 and stores it in the training data DB 104 in the storage unit 13. The processing unit 10A also acquires a second training parameter group from the blood cell counter 350 and stores it in the training data DB 104 in the storage unit 13. The processing unit 10A generates training data by linking the first training parameter group and the second training parameter group with disease name labels according to the method described in 1-2 above, and stores the data in the training data database 104 in the storage unit 13. The disease name labels corresponding to each parameter group may be received from the input unit 16 by the operator for each parameter group, and linked to the first training parameter group and the second training parameter group. Alternatively, when each parameter is acquired in the cell image analyzer 300 or the blood cell counter 350, the patient information may be linked to each parameter, and the processing unit 10A may read that information.

[0066] The processing unit 10A receives a training processing start command input by the operator from the input unit 16, and in step S12, reads out a computer algorithm stored in the algorithm DB 105 in the memory unit 13, and further reads out training data from the training data DB 10, and inputs the training data into the computer algorithm.

[0067] In step S13, the processing unit 10A determines whether the computer algorithm has been trained using all the training data, and if the computer algorithm has not been trained using all the training data (if "No"), it proceeds to step S14, reads the next training data from the training data DB10, and returns to step S12 to continue processing.

[0068] If the processing unit 10A has trained the computer algorithm using all the training data in step S13 (if "Yes"), the processing unit 10A proceeds to step S15 and records the trained computer algorithm in the algorithm DB 105 in the storage unit 13.

[0069] A trained computer algorithm acts as a discriminator that generates discrimination support information to aid in disease differentiation.

[0070] 2-4. Disease differentiation support device The disease differentiation support device 200A acquires a first group of analytical parameters, a second group of analytical parameters, and a discriminator, and generates differentiation support information for supporting disease differentiation. The disease differentiation support device 200A acquires the first analytical parameters from the cell image analyzer 400 via a recording medium 98 or a network 99. The disease differentiation support device 200A acquires the second analytical parameters from the blood cell counter 450 via the recording medium 98 or a network 99.

[0071] (1) Hardware configuration of the disease differentiation support device The hardware configuration of the disease differentiation support device 200A will be described with reference to Fig. 12. The configuration of the disease differentiation support device 200A is basically the same as that of the training device 100A. However, the processing unit 10 (10A), input unit 16, and output unit 17 of the training device 100A are to be read as the processing unit 20 (20A), input unit 26, and output unit 27 in the disease differentiation support device 200A.

[0072] In addition, the CPU 11, memory 12, storage unit 13, bus 14, interface unit 15, and GPU 19 of the training device 100A shall be read as a CPU 21, memory 22, storage unit 23, bus 24, interface unit 25, and GPU 29 in the disease differentiation support device 200A.

[0073] The processing unit 20 pre-records in the storage unit 13 a computer program for performing the processing of each step described in Fig. 14 below, for example, in an executable format. The executable format is, for example, a format generated by converting a programming language by a compiler. The processing unit 10 generates discrimination support information for supporting disease discrimination using the computer program for supporting disease discrimination recorded in the storage unit 13 and the discriminator generated by the training device 100A.

[0074] In addition, in order to perform the processing of each step described in the disease analysis processing below, the processing unit 20 pre-records a computer program and a classifier for supporting disease differentiation, which will be described later, in the storage unit 23, for example, in an executable format. The executable format is, for example, a format generated by conversion from a programming language by a compiler. The processing unit 20 cooperates with an operation system stored in the storage unit 23 to perform a process of generating discrimination support information for supporting disease differentiation using the program and the classifier for supporting disease differentiation.

[0075] In the following description, unless otherwise specified, the processing performed by the processing unit 20 means processing actually performed by the CPU 21 of the processing unit 20 based on a computer program and a classifier for assisting in disease discrimination stored in the storage unit 23 or memory 22. The CPU 21 uses the memory 22 as a working area to temporarily store necessary data (intermediate data during processing, etc.) in a volatile manner, and appropriately records data to be stored for a long time, such as calculation results, in the storage unit 23 in a non-volatile manner.

[0076] (2) Functional configuration of the disease differentiation support device 13, the processing unit 20A of the disease discrimination support device 200A functions as an analysis data acquisition unit 201, an analysis data input unit 202, an analysis unit 203, an analysis data database (DB) 204, and a classifier database (DB) 205. These functions are realized by installing a computer program (e.g., Python, etc.) that causes a computer to execute a process of generating discrimination support information in the storage unit 23 or memory 22 of the processing unit 20A, and executing this computer program and a computer program for supporting the discrimination of diseases including a classifier by the CPU 21. The analysis data database (DB) 204 stores a first analysis parameter group acquired by the processing unit 10A from the cell image analyzer 400 and a second analysis parameter group acquired from the blood cell counter 450. The classifier database (DB) 205 stores the classifier acquired from the training device 100.

[0077] The analysis data acquisition unit 201 corresponds to step S21, which will be described later, the analysis data input unit 202 corresponds to steps S22 and S23, and the analysis unit 203 corresponds to step S24.

[0078] (3) Processing of computer programs to assist in the differentiation of diseases The processing unit 20A of the disease differentiation support apparatus 200A executes each step shown in FIG.

[0079] Processing unit 20A receives an analysis data acquisition process start command input by the operator from input unit 26, and in step S21, acquires a first group of analysis parameters from cell image analyzer 400 and stores them in analysis data DB 204 in storage unit 23. Processing unit 20A also acquires a second group of analysis parameters from blood cell counter 450 and stores them in analysis data DB 204 in storage unit 23.

[0080] The processing unit 20A receives a classifier acquisition process start command input by the operator from the input unit 26, and acquires a classifier from the training device 100A in step S22. Alternatively, if the classifier is stored in advance in the classifier database 205 in the storage unit 23, the stored classifier is read out.

[0081] The processing unit 20A receives an analysis processing start command input by the operator from the input unit 26, and in step S23, calls up the first group of analysis parameters and the second group of analysis parameters acquired in step S21 from the analysis data DB 204 and inputs them to the classifier.

[0082] In step S24, the processing unit 20A generates a value indicating the probability of each disease as discrimination support information for supporting the discrimination of the disease, and records the value in the storage unit 23.

[0083] In step S25, the processing unit 20A outputs the result generated in step S24 to the output unit 27.

[0084] 3.Disease Differentiation Support System 2 Another aspect of the disease differentiation support system will be described with reference to Fig. 15 and Fig. 16. Fig. 15 shows a configuration example of the disease differentiation support system 2. The disease differentiation support system 2 includes a user side device 200, a cell image analyzer 400, and a blood cell counter 450, and the user side device 200 operates as a disease differentiation support device 200B that performs both training and disease differentiation support. The disease differentiation support device 200B functions as both the training device 100A and the disease differentiation support device 200A. The disease differentiation support device 200B is connected to the cell image analyzer 400 and the blood cell counter 450.

[0085] (1) Hardware configuration of the disease differentiation support device 200B The hardware configuration of the disease differentiation support apparatus 200B is similar to that of the user side apparatus 200 shown in Fig. 12. In Fig. 12, the processing unit 20A in the disease differentiation support apparatus 200B is to be read as the processing unit 20B.

[0086] (2) Functional configuration of the disease differentiation support device 200B FIG. 16 shows the functional configuration of the disease differentiation support device 200B. The processing unit 20B of the disease differentiation support device 200B functions as a training data generating unit 101, a training data input unit 102, an algorithm updating unit 103, an analysis data acquiring unit 201, an analysis data input unit 202, an analysis unit 203, a parameter database (DB) 304, and an algorithm database (DB) 305. The parameter database (DB) 304 has the functions of the training data DB 104 described in 2-3.(2) above and the analysis data DB 204 described in 2-4.(2) above. The algorithm database (DB) 305 has the functions of the algorithm DB 105 described in 2-3.(2) above and the classifier DB 205 described in 2-4.(2) above. That is, the parameter DB 304 stores the first training parameter group and the first analysis parameter group acquired by the processing unit 20B from the cell image analyzer 400, and the second training parameter group and the second analysis parameter group acquired from the blood cell counter 450. The algorithm DB 305 stores a classifier, which is a computer algorithm before training and a computer algorithm after training.

[0087] During training, the processing unit 20B of the disease differentiation support apparatus 200B performs the processes shown in 2-3.(3) above and in Fig. 11. The training data generating unit 101 corresponds to step S11 described in 2-3.(3) above, the training data input unit 102 corresponds to step S12, and the algorithm updating unit 103 corresponds to step S15. Here, the "storage unit 13 of the processing unit 10A," "training data DB 104," and "algorithm DB 105" described in 2-3.(3) above should be read as "storage unit 23 of the processing unit 20B," "parameter DB 304," and "algorithm DB 305," respectively.

[0088] The processing unit 20B of the disease differentiation support device 200B performs the processes shown in 2-4.(3) above and Fig. 14 during the generation process of the differentiation support information. The analysis data acquisition unit 201 corresponds to step S21 described in 2-4.(3) above, the analysis data input unit 202 corresponds to steps S22 and S23, and the analysis unit 203 corresponds to step S24. Here, the "analysis data DB204" and the "classifier DB205" described in 2-4.(3) above are to be read as "parameter DB304" and "algorithm DB305", respectively.

[0089] 4.Disease Differentiation Support System 3 Another aspect of the disease differentiation support system will be described with reference to Fig. 17 and Fig. 18. Fig. 17 shows a configuration example of the disease differentiation support system 3. The disease differentiation support system 3 includes a vendor-side device 100 and a user-side device 200. The vendor-side device 100 includes a processing unit 10 (10B), an input unit 16, and an output unit 17. The vendor-side device 100 operates as a disease differentiation support device 100B that performs both training processing and differentiation support information generation processing, similar to the disease differentiation support device 200B. On the other hand, the user-side device 200 operates as a terminal device 200C.

[0090] Here, the disease differentiation support device 100B is a device on the cloud server side composed of, for example, a general-purpose computer, etc. The disease differentiation support device 100B is communicably connected to a cell image analyzer 300 and a blood cell counter 350. In addition, the disease differentiation support device 100B is communicably connected to a terminal device 200C via a network 99. The terminal device 200C is a general-purpose computer or the like, and is connected to the cell image analyzer 400 and the blood cell counter 450 so as to be able to communicate with each other.

[0091] (1) Hardware configuration of the disease differentiation support device 100B The hardware configuration of the disease differentiation support apparatus 100B is similar to that of the vendor-side apparatus 100 shown in Fig. 10. In Fig. 10, the processing unit 10A is to be read as a processing unit 10B in the disease differentiation support apparatus 100B.

[0092] (2) Functional configuration of disease differentiation support system 3 18 shows the functional configuration of the disease differentiation support system 3. The functional configuration of the disease differentiation support device 100B is similar to that in 3.(2) and FIG.

[0093] The terminal device 200C acquires first analysis parameters from the cell image analyzer 400 and second analysis parameters from the blood cell counter 450, and transmits these analysis parameters to the disease differentiation support device 100B via the network 99. The disease differentiation support device 100B generates differentiation support information from the analysis parameters transmitted from the terminal device 200C, and transmits the generated information to the terminal device 200C.

[0094] 5. Computer Programs Another embodiment of the present disclosure relates to a computer program that causes a computer to execute a training process including steps S11 to S15 shown in 2-3.(3) above and in FIG.

[0095] Another embodiment of the present disclosure relates to a computer program that causes a computer to execute a process for supporting disease differentiation, including steps S21 to S25 shown in 2-4.(3) above and FIG.

[0096] The computer program may be provided as a program product, such as a recording medium, on which the computer program is stored. The computer program is stored in a recording medium, such as a hard disk, a semiconductor memory element such as a flash memory, or an optical disk. The format of the program stored in the recording medium is not limited as long as the processing unit can read the program. It is preferable that the storage in the recording medium is non-volatile.

[0097] 6.Verification of effectiveness The effectiveness of the method for assisting in disease differentiation disclosed in this specification was verified.

[0098] 6-1. Sample (1) Training samples Between February and September 2017, peripheral blood samples were collected using EDTA from 34 patients with polycythemia vera (PV), 168 patients with essential thrombocythemia (ET), and 69 patients with primary myelofibrosis (PMF) who had been diagnosed and visited Juntendo University Hospital. These samples were used as training samples.

[0099] (2) Verification samples Peripheral blood samples were collected using EDTA from 9 patients with polycythemia vera (PV), 53 patients with essential thrombocythemia (ET), and 12 patients with primary myelofibrosis (PMF) who had already been diagnosed at Juntendo University Hospital and used as validation samples.

[0100] 6-2. Acquisition and selection of each parameter (1) Obtaining the first set of parameters Peripheral blood smears were prepared and parameters related to abnormal findings were obtained using an Automated Digital Cell Morphology Analyzer DI-60 (Sysmex Corporation). Abnormal findings were obtained for a total of 181 items, including 17 cell types and classification values ​​and 164 abnormal morphological characteristics.

[0101] (2) Obtaining the second set of parameters Using the XN series blood cell counter, measurements of 174 blood cell test items were obtained.

[0102] (3) Selection of the first and second parameters In order to select items from the first and second parameter groups that are more closely related to disease differentiation, one-dimensional analysis of variance (ANOVA) was used for selection.

[0103] The null hypothesis H0 of the ANOVA was "there is no difference between the groups," and the alternative hypothesis H1 was "there is a difference between the groups." The significance level p-value was set at 0.05, and items with p<0.05 were extracted as items highly related to the disease. 44 items were extracted from the first parameter group and 121 items from the second parameter group using the significance test by ANOVA. Examples of the extracted items and p-values ​​are shown in Table 1.

[0104] [Table 1]

[0105] 6-3. Training For each training sample, the first and second parameter groups selected in 6-2.(3) above were obtained, and a matrix in which these were arranged in the same hierarchy was generated for each training sample. Training data was created by linking the disease labels of each patient from whom the training sample was drawn to each matrix. The created training data was input into a gradient boosting tree to train the algorithm and generate a classifier. Python was used as the software.

[0106] Verification For each verification sample, the first and second parameter groups selected in 6-2.(3) above were obtained, and a matrix in which these were arranged in the same hierarchy was generated for each verification sample, and this was used as the analysis data for each verification sample. Each generated analysis data was input into the discriminator, and the discrimination results were obtained.

[0107] Figure 19 shows a comparison of the results of the machine method using the discriminator and the definitive diagnosis by a doctor. Of the nine patients diagnosed with PV by a doctor, all were predicted to be PV by the machine method. Of the 53 patients diagnosed with ET by a doctor, 49 were also predicted to be ET by the machine method, two were predicted to be PV by the machine method, and two were predicted to be PMF. In addition, of the 12 patients diagnosed with PMF by a doctor, 10 were also predicted to be PMF by the machine method, one was predicted to be PV, and one was predicted to be ET.

[0108] Figures 20, 21, and 22 show the ROC curves for PV, ET, and PMF predicted by the machine method, respectively. Figure 23 shows the sensitivity, specificity, and AUC values ​​obtained from each ROC curve. For all diseases, the sensitivity and specificity were good, exceeding 90%. The AUC values ​​were also good, exceeding 0.96.

[0109] 6-5. Comparison with conventional methods Figure 24 shows Figure B of Non-Patent Document 1. In the method of Non-Patent Document 1, there are many cases where the PMF is judged as ET or PT, or where it cannot be judged. For this reason, it was considered that the method of supporting disease differentiation disclosed in this specification is more suitable for supporting disease differentiation. [Explanation of symbols]

[0110] 200A,200B,100B Disease differentiation support device 20A, 20B, 10B Processing section

Claims

1. A disease differentiation support method for supporting disease differentiation, comprising: obtaining a first parameter related to the number of cells having abnormal morphology obtained by analyzing an image of cells contained in a sample taken from the subject; obtaining a second parameter related to the number of each type of cell in the sample; generating, using a pre-trained computer algorithm, differentiation support information for supporting a disease differentiation based on the first parameter and the second parameter; Obtaining the first parameter includes obtaining the first parameter based on analysis by a first cell image analysis device that analyzes the image; obtaining the second parameter includes obtaining the second parameter based on an analysis by a first blood cell counter that analyzes an optical signal or an electrical signal obtained from cells contained in the sample; the computer algorithm is pre-trained based on first training parameters selected from a plurality of parameters acquired based on the analysis of the second cell image analyzer, second training parameters selected from a plurality of parameters acquired based on the analysis of the second blood cell counter, and a disease label representing a disease; the first training parameters are parameters selected based on a result of a significant difference test between disease groups for each of the plurality of parameters acquired based on the analysis by the second cell image analyzer, the second training parameters are parameters selected based on a result of a significant difference test between the disease groups for each of the plurality of parameters acquired based on an analysis of the second blood cell counter, the first parameter and the training first parameter are parameters of the same type, the second parameter and the training second parameter are parameters of the same type, The first cell-image analysis device and the second cell-image analysis device are the same or different devices of the same type, The first blood cell counting device and the second blood cell counting device are the same or different devices of the same type. Disease differentiation support method.

2. The first cell image analysis device analyzes the image captured by an imaging unit. The disease differentiation support method according to claim 1.

3. The first blood cell counter analyzes the optical signal or the electrical signal obtained by the detection unit. The disease differentiation support method according to claim 1 or 2.

4. the detection unit includes a flow cytometer; The first blood cell counter analyzes the optical signal obtained by the flow cytometer. The disease differentiation support method according to claim 3.

5. The abnormal morphology includes at least one selected from nuclear morphology abnormality, granular abnormality, cell size abnormality, cell malformation, cell destruction, vacuoles, immature cells, the presence of inclusion bodies, Dohle bodies, satellite phenomenon, nuclear network abnormality, petal-like nuclei, large N / C ratio, bleb-like morphology, smudge, and hairy cell-like morphology. The disease differentiation support method according to claim 1.

6. The nuclear morphological abnormality includes at least one selected from hyper-lobulation, hypo-lobulation, pseudo-Pelger anomaly, ring-shaped nucleus, spherical nucleus, oval nucleus, apoptosis, multinuclei, karyorrhexis, enucleation, bare nucleus, irregular nuclear margin, nuclear fragmentation, internuclear bridges, multiple nuclei, cleaved nuclei, nuclear division, and nucleolar abnormalities; The granular abnormality includes at least one selected from degranulation, granule distribution abnormality, toxic granules, Auer rods, Bassoon cells, and pseudo-Chediak-Higashi granule-like granules; The cell size abnormality includes giant platelets. The disease differentiation support method according to claim 5.

7. The first parameter includes a parameter related to at least one of the number and ratio of atypical lymphocytes; or a parameter related to at least one of the number of each type of cell having at least one abnormal morphology selected from neutrophils, eosinophils, platelets, lymphocytes, monocytes, basophils, metamyelocytes, myelocytes, promyelocytes, blasts, plasma cells, immature eosinophils, immature basophils, erythroblasts, and megakaryocytes, and the ratio of each type of cell having the abnormal morphology. The disease differentiation support method according to any one of claims 1 to 6.

8. The second parameter includes a parameter related to the number of each type of cell or the ratio of each type of cell. The disease differentiation support method according to claim 1 .

9. The second parameter includes at least one parameter related to the number of at least one type of cell selected from red blood cells, nucleated red blood cells, small red blood cells, platelets, reticulocytes, immature granulocytes, neutrophils, eosinophils, basophils, lymphocytes, and monocytes, and the ratio of each type of cell. The disease differentiation support method according to any one of claims 1 to 8.

10. the computer algorithm comprises a machine learning algorithm; The disease differentiation support method according to any one of claims 1 to 9.

11. The machine learning algorithm comprises an algorithm selected from trees, regression, support vector machines, Bayes, clustering, and random forests. The disease differentiation support method according to claim 10.

12. The machine learning algorithm comprises a gradient boosting tree algorithm. The disease differentiation support method according to claim 11.

13. Obtaining the first parameter includes analyzing the image using a deep learning algorithm having a neural network structure. The disease differentiation support method according to any one of claims 1 to 12.

14. Acquiring the first parameter includes acquiring the first parameter based on a classification result of the cell output from the deep learning algorithm and a probability of being classified into the cell. The disease differentiation support method according to claim 13.

15. The disease is a hematopoietic disease. The disease differentiation support method according to any one of claims 1 to 14.

16. The hematopoietic disease is a myeloproliferative neoplasm. The disease differentiation support method according to claim 15.

17. The myeloproliferative neoplasm comprises polycythemia vera, essential thrombocythemia, or primary myelofibrosis. The disease differentiation support method according to claim 16.

18. The hematopoietic disease is leukemia, myelodysplastic syndrome, lymphoma, or myeloma; The disease differentiation support method according to claim 15.

19. A disease differentiation support device for supporting disease differentiation, comprising: a processor for acquiring a first parameter related to the number of cells having an abnormal morphology obtained by analyzing an image of cells contained in a sample collected from a subject, acquiring a second parameter related to the number of each type of cell contained in the sample, and generating, using a pre-trained computer algorithm, differentiation support information for supporting disease differentiation based on the first parameter and the second parameter; the processing unit acquires the first parameter based on an analysis by a first cell image analysis device that analyzes the image, and acquires the second parameter based on an analysis by a first blood cell counting device that analyzes an optical signal or an electrical signal obtained from a cell contained in the sample; the computer algorithm is pre-trained based on first training parameters selected from a plurality of parameters acquired based on the analysis of the second cell image analyzer, second training parameters selected from a plurality of parameters acquired based on the analysis of the second blood cell counter, and a disease label representing a disease; the first training parameters are parameters selected based on a result of a significant difference test between disease groups for each of the plurality of parameters acquired based on the analysis by the second cell image analyzer, the second training parameters are parameters selected based on a result of a significant difference test between the disease groups for each of the plurality of parameters acquired based on an analysis of the second blood cell counter, the first parameter and the training first parameter are parameters of the same type, the second parameter and the training second parameter are parameters of the same type, The first cell-image analysis device and the second cell-image analysis device are the same or different devices of the same type, The first blood cell counting device and the second blood cell counting device are the same or different devices of the same type. Disease differentiation support device.

20. On the computer, obtaining a first parameter related to the number of cells having abnormal morphology obtained by analyzing an image of cells contained in a sample taken from the subject; obtaining a second parameter related to the number of each type of cell in the sample; generating, using a pre-trained computer algorithm, differentiation support information for supporting a disease differentiation based on the first parameter and the second parameter; Run the command, The step of acquiring the first parameter includes a step of acquiring the first parameter based on an analysis by a first cell image analysis device that analyzes the image, the step of acquiring the second parameter includes a step of acquiring the second parameter based on an analysis by a first blood cell counter that analyzes an optical signal or an electrical signal obtained from cells contained in the sample; the computer algorithm is pre-trained based on first training parameters selected from a plurality of parameters acquired based on the analysis of the second cell image analyzer, second training parameters selected from a plurality of parameters acquired based on the analysis of the second blood cell counter, and a disease label representing a disease; the first training parameters are parameters selected based on a result of a significant difference test between disease groups for each of the plurality of parameters acquired based on the analysis by the second cell image analyzer, the second training parameters are parameters selected based on a result of a significant difference test between the disease groups for each of the plurality of parameters acquired based on an analysis of the second blood cell counter, the first parameter and the training first parameter are parameters of the same type, the second parameter and the training second parameter are parameters of the same type, The first cell-image analysis device and the second cell-image analysis device are the same or different devices of the same type, The first blood cell counting device and the second blood cell counting device are the same or different devices of the same type. Computer program.

Images