Cell analysis method and cell analysis apparatus

Abstract

To provide a cell analysis method and a cell analyzer that analyze data obtained from a plurality of cells included in a specimen, and can process data of cells with a largely increased information amount with required throughput.SOLUTION: A cell analyzer comprises: measurement units (400, 400a, 500, 500a, 700) that each measure a plurality of cells included in a sample; first processors (3001, 4831, 6001, 8111) that each perform information processing related to analysis of the cells; and second processors (3002, 4833, 6002, 8112) that each perform parallel processing. The measurement unit acquires matrix data including, as elements, values indicating an analog signal level corresponding to the intensities of light emitted from the cells at a plurality of time points of each of the cells in a flow cytometer. The second processor executes the parallel processing for processing the matrix data according to an artificial intelligence algorithm including a matrix operation. The first processor classifies the respective cell classes of the cells on the basis of a result of the parallel processing.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a cell analysis method and a cell analysis apparatus.

Background Art

[0002] Patent Document 1 describes a method of analyzing data obtained by measuring blood cells with a flow cytometer in a data processing system equipped with a processor and classifying cells according to their types. In Patent Document 1, it is described that when cells cannot be classified by the optical information measured by the flow cytometer, information on the volume and conductivity of the cells is further used.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When increasing the amount of information used for cell classification in order to improve the classification accuracy of cells, in a specimen such as blood or urine containing a plurality of cells, the data volume per specimen becomes enormous due to an increase in the amount of information obtained from one cell. For example, in order to classify individual cells using a deep learning algorithm, it is necessary to significantly increase the amount of information obtained from each cell in order to extract features for each cell. Patent Document 1 does not disclose a system capable of processing the significantly increased amount of information within the required processing capacity.

[0005] One aspect of the present invention is to provide a cell analysis method and a cell analysis apparatus capable of satisfying the requirements of the required processing capacity for cell data with a significantly increasing amount of information in a configuration for analyzing data obtained from a plurality of cells contained in a specimen.

Means for Solving the Problems

[0006] To solve the above problems, a cell analyzer according to one aspect of the present invention includes a flow cytometer and comprises a measurement unit (400, 400a, 500, 500a, 700) configured to measure a plurality of cells contained in a sample, a first processor (3001, 4831, 6001, 8111) configured to perform information processing related to the analysis of the plurality of cells, and a second processor (3002, 4833, 6002, 8112) configured to perform parallel processing, wherein the measurement unit (400, 400a, 500, 500a, 700) flows the sample through the flow cell of the flow cytometer and releases light from the irradiated cells. By acquiring analog signals corresponding to the intensity of the light being emitted for each of multiple cells, matrix data is obtained whose elements are digital values ​​representing the analog signal levels of each of the multiple cells at multiple time points. The second processors (3002, 4833, 6002, 8112) are configured to perform parallel processing to process the matrix data according to an artificial intelligence algorithm that includes multiple matrix operations. The first processors (3001, 4831, 6001, 8111) are configured to classify the cell type of each of the multiple cells based on the results of the parallel processing by the second processors (3002, 4833, 6002, 8112).

[0007] To solve the above problems, another aspect of the present invention provides a cell analysis method using a cell analyzer including a first processor (3001, 4831, 6001, 8111), a second processor (3002, 4833, 6002, 8112), and a flow cytometer, comprising: flowing a sample through the flow cell of the flow cytometer and acquiring an analog signal corresponding to the intensity of light emitted from the irradiated cells for each of the plurality of cells, thereby acquiring matrix data for each of the plurality of cells in the sample, wherein the matrix data has as elements values ​​that digitally represent the analog signal levels at multiple time points, the second processor performs parallel processing to process the matrix data according to an artificial intelligence algorithm that includes multiple matrix operations, and the first processor classifies the cell type of each of the plurality of cells based on the results of the parallel processing by the second processor. [Effects of the Invention]

[0008] This makes it possible to analyze multiple cells based on large amounts of data while meeting the processing power requirements. [Brief explanation of the drawing]

[0009] [Figure 1] (a) shows an example of the conventional method of leukocyte classification. (b) shows an example of the method of leukocyte classification. [Figure 2] (a) shows an example of irradiating cells flowing through a flow cell with light. (b) shows an example of sampling forward scattered light signals, side scattered light signals, and fluorescence signals. (c) shows an example of waveform data obtained by sampling. [Figure 3] An example of how to generate training data is shown. [Figure 4] Here are some examples of label values. [Figure 5] Here are some examples of methods for analyzing analytical data. [Figure 6] An example of the appearance of a cell analysis device is shown. [Figure 7] An example of a block diagram of a measurement unit is shown. [Figure 8] An example of a specimen suction unit and a sample preparation unit is shown. [Figure 9] An example of the configuration of the optical system of the FCM detection unit is shown. [Figure 10] An example of the configuration of the processing unit is shown. [Figure 11] An example of the configuration of the parallel processing processor is shown. [Figure 12] An example of the implementation of the parallel processing processor in the measurement unit is shown. [Figure 13] Another example of the implementation of the parallel processing processor in the measurement unit is shown. [Figure 14] Another example of the implementation of the parallel processing processor in the measurement unit is shown. [Figure 15] Another example of the implementation of the parallel processing processor in the measurement unit is shown. [Figure 16] An overview of the operation in which the processor executes arithmetic processing of matrix data using the parallel processing processor is shown. [Figure 17] (a) shows the calculation formula for the product of matrices. (b) shows an example of the arithmetic processing executed in parallel by the parallel processing processor. [Figure 18] The state of executing the arithmetic processing by the parallel processing processor is shown. [Figure 19] (a) shows an example of the waveform data of the forward scattered light as the waveform data input to the deep learning algorithm. (b) shows an overview of the matrix operation between the waveform data and the filter. [Figure 20] An example of the analysis operation of the specimen by the cell analyzer is shown. [Figure 21] An example of the cell analysis process is shown. [Figure 22] An example of parallel processing is shown. [Figure 23] Another example of the block diagram of the measurement unit is shown. [Figure 24] An example of the block diagram of the processing unit is shown. [Figure 25] An overview of the operation in which the processor executes arithmetic processing of matrix data using the parallel processing processor is shown. [Figure 26] Another example of the block diagram of the measurement unit is shown. [Figure 27] Here are some other examples of block diagrams for processing units. [Figure 28] An example configuration of a measurement unit, processing unit, and analysis unit is shown. [Figure 29] Another example of a block diagram for a measurement unit is shown. [Figure 30] An example of a block diagram for an analysis unit is shown. [Figure 31] This outlines how a processor uses parallel processing processors to perform arithmetic operations on matrix data. [Figure 32] Here are some other examples of block diagrams for processing units. [Figure 33] Another example of a block diagram for a measurement unit is shown. [Figure 34] Here are some other examples of block diagrams for analysis units. [Figure 35] An example of a block diagram of a measurement unit is shown. [Figure 36] A schematic example of the optical system for a flow cytometer is shown. [Figure 37] A schematic example of the sample preparation section of the measurement unit is shown. [Figure 38] A schematic example of a waveform data analysis system is shown. [Figure 39] An example of a block diagram of the vendor's equipment is shown. [Figure 40] An example of a block diagram of a measurement unit is shown. [Figure 41] Here are some other examples of block diagrams for analysis units. [Figure 42] An example of a functional block diagram for a deep learning system is shown. [Figure 43] An example flowchart of the processing unit's operation for generating training data is shown. [Figure 44] The following are schematic diagrams illustrating neural networks. (a) shows a schematic diagram illustrating the overview of a neural network. (b) shows a schematic diagram illustrating the operations at each node. (c) shows a schematic diagram illustrating the operations between nodes. [Figure 45]A mixed matrix of the results obtained using the reference method and the results obtained using the deep learning algorithm is shown. [Figure 46] (a) shows the ROC curve for neutrophils. (b) shows the ROC curve for lymphocytes. (c) shows the ROC curve for monocytes. [Figure 47] (a) shows the ROC curve for eosinophils. (b) shows the ROC curve for basophils. (c) shows the ROC curve for control blood (CONT). [Figure 48] This shows an example configuration of a cell analysis device used as an image analysis device. [Figure 49] An example of the processing unit configuration is shown. [Figure 50] An example of how to generate training data is shown. [Figure 51] Here are some examples of label values. [Figure 52] Here are some examples of image analysis methods. [Figure 53] One embodiment of the analysis results is shown. [Modes for carrying out the invention]

[0010] The outline and embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In the following description and drawings, the same reference numerals indicate the same or similar components, and therefore, descriptions of the same or similar components will be omitted.

[0011] [1. Methods for analyzing cells] This embodiment discloses a cell analysis method in a cell analysis device including a host processor and a parallel processing processor, which includes acquiring data about each cell based on control by the host processor, performing parallel processing on the data with the parallel processing processor, and generating cell type information for each of the cells based on the results of the parallel processing.

[0012] This analysis method allows for the parallel processing of cell data using a separate parallel processing processor, even when analyzing massive amounts of data ranging from several hundred megabytes to several gigabytes per sample. Therefore, even when classifying cells using a deep learning algorithm with enormous amounts of data, the data processing can be completed solely within the cell analyzer. For example, there is no need to transmit cell data to an analysis server containing the deep learning algorithm via the internet or intranet. Consequently, this analysis method eliminates the need to transmit large amounts of data from the cell analyzer to an analysis server and retrieve analysis results from the server, thus improving cell classification accuracy while maintaining the processing power of the cell analyzer.

[0013] An example of the outline of this embodiment will be explained using Figure 1. Figure 1(a) is a schematic representation of the leukocyte classification of the conventional method, and Figure 1(b) is a schematic representation of the leukocyte classification of the present method. In Figures 1(a) and 1(b), FSC represents an analog signal indicating the signal intensity of forward scattered light, SSC represents an analog signal indicating the signal intensity of side scattered light, and SFL represents an analog signal indicating the signal intensity of side fluorescence. As shown in Figure 1(a), in the conventional method, individual cells contained in the sample are measured with a flow cytometer, and the peak heights of the pulses of the analog signals of forward scattered light, side scattered light, and side fluorescence are obtained as the forward scattered light intensity, side scattered light intensity, and side fluorescence intensity, respectively. Next, the cells are classified into specific types based on the forward scattered light intensity, side scattered light intensity, and side fluorescence intensity. The result of the cell classification is displayed as a scattergram as shown in Figure 1(a). In the scattergram of Figure 1(a), the horizontal axis represents the intensity of side scattered light, and the vertical axis represents the intensity of side fluorescence.

[0014] Conventional leukocyte classification, as shown in Figure 1(a), determined the type of blood cell based solely on information about the peak height of the analog waveform. In contrast, the method of this embodiment classifies cells by analyzing the entire waveform of the analog signal acquired from a single cell by a flow cytometer as the data to be analyzed, as shown in Figure 1(b). Figure 1(b) shows the waveform plotted from the analog signal obtained by the flow cytometer, but as will be described later, in this embodiment, the data concerning cells in the sample is intended to be digital data (waveform data, described later) whose elements are values ​​indicating the signal intensity at multiple time points obtained by A / D conversion of this analog signal. This group of digital data is matrix data, and in this embodiment, for example, it is matrix data with one row and multiple columns (i.e., one-dimensional array data).

[0015] In this embodiment, the deep learning algorithm 50 shown in Figure 1(b) is trained with waveform data for each cell type. Then, the waveform data of cells of unknown cell type contained in the sample is input to the trained deep learning algorithm 60, and the deep learning algorithm 60 derives a cell type determination result for each cell. Deep learning algorithms 50 and 60 are artificial intelligence algorithms and consist of neural networks including multiple hidden layers. In this embodiment, when performing processing related to the analysis of waveform data according to the trained deep learning algorithm 60, the matrix operations that are included in large quantities in the deep learning algorithm 60 are executed in parallel using a parallel processing processor installed in the cell analyzer. The cell analyzer comprises a parallel processing processor capable of executing parallel processing and an execution instruction processor (hereinafter simply referred to as a processor) that causes the parallel processing processor to execute parallel processing.

[0016] Hereinafter, individual cells in a biological sample subjected to analysis for the purpose of determining cell type will also be referred to as "cells to be analyzed." In other words, a biological sample may contain multiple cells to be analyzed. Multiple cells may contain multiple types of cells to be analyzed.

[0017] Examples of biological samples include biological samples collected from a subject. For example, biological samples 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 fluid, cerebrospinal fluid, etc. Hereinafter, body fluids other than blood and urine may simply be referred to as "body fluids." Blood samples are not limited as long as they are in a state where cell counting and cell type determination can be made. Blood is preferably peripheral blood. For example, blood may be peripheral blood collected using an anticoagulant such as ethylenediaminetetraacetate sodium salt or potassium salt, or heparin sodium. Peripheral blood may be collected from an artery or a vein.

[0018] The cell types to be determined in this embodiment are 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 person, the cell types to be determined in this embodiment include, for example, nucleated cells such as red blood cells and white blood cells, and platelets. Nucleated cells include, for example, neutrophils, lymphocytes, monocytes, eosinophils, and basophils. Neutrophils include, for example, segmented neutrophils and band neutrophils. On the other hand, when the blood is collected from a non-healthy person, nucleated cells may include, for example, at least one selected from the group consisting of immature granulocytes and abnormal cells. Such cells are also included in the cell types to be determined in this embodiment. Immature granulocytes may include, for example, metamyelocytes, myelocytes, promyelocytes, myeloblasts, and other cells.

[0019] Furthermore, nucleated cells may include not only normal cells but also abnormal cells not found in the peripheral blood of healthy individuals. Examples of abnormal cells are cells that appear when a person suffers from a specified disease, such as tumor cells. In the case of the hematopoietic system, the specified disease may be selected from the group consisting of, for example, 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 lymphoblastic leukemia, lymphoblastic leukemia, chronic myeloid leukemia, or chronic lymphoblastic leukemia, malignant lymphomas such as Hodgkin lymphoma and non-Hodgkin lymphoma, and multiple myeloma.

[0020] Furthermore, abnormal cells may include, for example, erythroblasts, which are nucleated red blood cells such as lymphoblasts, plasma cells, atypical lymphocytes, reactive lymphocytes, proerythroblasts, basophilic erythroblasts, polychromatic erythroblasts, orthochromatic erythroblasts, promegaloblasts, basophilic megaloblasts, polychromatic megaloblasts, and orthochromatic megaloblasts, as well as megakaryocytes, which include micromegacaryocytes, and other cells not normally found in the peripheral blood of healthy individuals.

[0021] Furthermore, if the biological sample is urine, the cell types to be determined in this embodiment may include, for example, red blood cells, white blood cells, transitional epithelium, squamous epithelium and other epithelial cells. Abnormal cells may include, for example, bacteria, filamentous fungi, yeast and other fungi, tumor cells, and the like.

[0022] When a biological sample is a bodily fluid that does not normally contain blood components, such as ascites, pleural fluid, or cerebrospinal fluid, the cell types may include, for example, red blood cells, white blood cells, and large cells. Here, "large cells" refers to cells that are larger than white blood cells and have been detached from the endometrium of the body cavity or the peritoneum of internal organs, such as mesothelial cells, histiocytes, and tumor cells.

[0023] When the biological sample is bone marrow fluid, the cell types to be determined in this embodiment may include mature hematopoietic cells and immature hematopoietic cells as normal cells. Mature hematopoietic 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 neutrophils. Immature hematopoietic cells include, for example, hematopoietic stem cells, immature granulocytes, immature lymphocytes, immature monocytes, immature red blood cells, megakaryocytes, and mesenchymal cells. Immature granulocytes may include, for example, metamyelocytes, myelocytes, promyelocytes, and myeloblasts. Immature lymphocytes include, for example, lymphoblasts. Immature monocytes include monoblasts. Immature erythroid cells include, for example, nucleated erythrocytes such as proerythroblasts, basophilic erythroblasts, polychromatic erythroblasts, orthochromatic erythroblasts, promegaloblasts, basophilic megaloblasts, polychromatic megaloblasts, and orthochromatic megaloblasts. Megakaryocytic cells include, for example, megakaryoblasts.

[0024] Abnormal cells that may be found in the bone marrow include, for example, hematopoietic tumor cells selected from the group consisting of leukemias such as myelodysplastic syndrome, acute myeloblastic leukemia, acute promyelocytic leukemia, acute myelomonocytic leukemia, acute monocytic leukemia, erythroleukemia, acute megakaryoblastic leukemia, acute myeloid leukemia, acute lymphoblastic leukemia, lymphoblastic leukemia, chronic myeloid leukemia, or chronic lymphoblastic leukemia, malignant lymphomas such as Hodgkin lymphoma and non-Hodgkin lymphoma, and multiple myeloma, as well as metastatic tumor cells of malignant tumors that have occurred in organs other than the bone marrow.

[0025] Figure 1 illustrates forward scatter, side scatter, and side fluorescence signals as examples of signals obtained from cells, which are optical signals obtained by irradiating cells flowing through a flow cell with light. However, there are no particular limitations as long as the signals represent the characteristics of the cells and can be used to classify cells by type.

[0026] The signal obtained from the cell may be any of the following: a signal representing the morphological characteristics of the cell, a signal representing the chemical characteristics, a signal representing the physical characteristics, or a signal representing the genetic characteristics of the cell, but preferably a signal representing the morphological characteristics of the cell. Preferably, the signal representing the morphological characteristics of the cell is an optical signal obtained from the cell.

[0027] The optical signal is preferably a light signal obtained as an optical response by irradiating a cell with light. The optical signal may include at least one selected from signals based on light scattering, signals based on light absorption, signals based on transmitted light, and signals based on fluorescence.

[0028] Signals based on light scattering may include scattered light signals and light loss signals resulting from light irradiation. The scattered light signals represent different parameters that characterize cells, depending on the angle at which the scattered light is received relative to the direction of light propagation. Forward scattered light signals are used as parameters representing cell size. Lateral scattered light signals are used as parameters representing the complexity of the cell nucleus.

[0029] In forward-scattered light, "forward" refers to the direction of propagation of light emitted from the light source. "Forward" may include forward low angles where the receiving angle is approximately 0 to 5 degrees, and / or forward high angles where the receiving angle is approximately 5 to 20 degrees, with the angle of the irradiated light being 0 degrees. "Sideways" is not limited as long as it does not overlap with "forward". "Sideways" may include receiving angles approximately 25 to 155 degrees, preferably 45 to 135 degrees, and more preferably 90 degrees, with the angle of the irradiated light being 0 degrees.

[0030] Signals based on light scattering may include polarization or depolarization as signal components. For example, by irradiating cells with light and receiving the scattered light through a polarizer, only scattered light polarized at a specific angle can be received. Alternatively, by irradiating cells with light through a polarizer and receiving the resulting scattered light through a polarizer that transmits only polarization at a different angle than the irradiating polarizer, only depolarized scattered light can be received.

[0031] The light loss signal represents the amount of light loss, based on the decrease in the amount of light received at the light-receiving part due to scattering of light when it is irradiated onto a cell. Preferably, the light loss signal is obtained as light loss in the optical axis direction of the irradiated light (axial light loss). The light loss signal can be expressed as the ratio of the amount of light received when the cell is flowing through the flow cell to the amount of light received when the cell is flowing through the flow cell, with the amount of light received at the light-receiving part when the cell is not flowing through the flow cell being set to 100%. Although axial light loss is used as a parameter representing the size of the cell, similar to the forward scattered light signal, the signal obtained will differ depending on whether the cell is translucent or not.

[0032] The fluorescence-based signal may be fluorescence excited by irradiating cells labeled with a fluorescent substance with light, or it may be autofluorescence generated from unstained cells. The fluorescent substance may be a fluorescent dye that binds to nucleic acids or membrane proteins, or it may be a labeled antibody that is modified with a fluorescent dye to bind to a specific protein in a cell.

[0033] Optical signals may be acquired in the form of image data obtained by irradiating cells with light and imaging the irradiated cells. Image data can be obtained by imaging individual cells flowing through a channel using an image sensor such as a TDI camera or CCD camera with a so-called imaging flow cytometer. Alternatively, image data of cells may be obtained by spreading, scattering, or dotting a sample or measurement sample containing cells onto a glass slide and imaging the glass slide with an image sensor.

[0034] The signals obtained from cells are not limited to optical signals; they may also be electrical signals. For example, the electrical signal may be obtained by applying a direct current to a flow cell and using the change in impedance caused by the cells flowing through the flow cell as the electrical signal. The electrical signal obtained in this way will be a parameter that reflects the volume of the cell. Alternatively, the electrical signal may be obtained by applying a radio frequency to cells flowing through a flow cell and using the change in impedance as the electrical signal. The electrical signal obtained in this way will be a parameter that reflects the conductivity of the cell.

[0035] The signals obtained from cells may be a combination of at least two of the signals obtained from cells as described above. By combining multiple signals, the characteristics of cells can be analyzed from multiple perspectives, enabling more accurate cell classification. The combination may be, for example, a combination of at least two of multiple optical signals, such as forward scattered light signals, side scattered light signals, and fluorescence signals, or a combination of scattered light signals at different angles, such as a low-angle scattered light signal and a high-angle scattered light signal. Alternatively, optical signals and electrical signals may be combined, and there are no particular restrictions on the type and number of signals to be combined.

[0036] In this embodiment, the method for analyzing cells is not limited to determining the cell type using a deep learning algorithm. For each cell passing through a predetermined location in a flow channel, the signal intensity may be acquired at multiple time points while the cell is passing through the predetermined location, and the cell type may be determined based on the recognition of the signal intensity at multiple time points for each cell as a pattern. The pattern may be recognized as a numerical pattern of signal intensity at multiple time points, or as a shape pattern when the signal intensity at multiple time points is plotted as a graph. When recognized as a numerical pattern, the cell type can be determined by comparing the numerical pattern of the cell under analysis with a numerical pattern of a cell type that is already known. For example, Spearman's rank correlation, z-score, etc., can be used to compare the numerical pattern of the cell under analysis with a control numerical pattern. The cell type can be determined by comparing the graph shape pattern of the cell under analysis with a graph shape pattern of a cell type that is already known. The graph shape patterns of the cells under analysis can be compared with graph shape patterns of cells whose cell types are already known, for example, using geometric shape pattern matching or feature descriptors such as SIFT Descriptor.

[0037] <Overview of cell analysis methods> Next, we will explain the method for generating training data 75 and the method for analyzing waveform data using the examples shown in Figures 2, 3 to 5.

[0038] <Waveform data> Figure 2 is a schematic diagram illustrating the waveform data used in this analysis method. As shown in Figure 2(a), when a sample containing cells C is flowed through a flow cell FC and light is shone on the cells C flowing through the flow cell FC, forward scattered light (FSC) is generated in front of the direction of light propagation. Similarly, side scattered light (SSC) and side fluorescence (SFL) are generated to the sides of the direction of light propagation. The forward scattered light is received by the first light-receiving unit D1, and a signal corresponding to the amount of light received is output. The side scattered light is received by the second light-receiving unit D2, and a signal corresponding to the amount of light received is output. The side fluorescence is received by the third light-receiving unit D3, and a signal corresponding to the amount of light received is output. As a result, analog signals representing the change in the signal over time are output from the light-receiving units D1 to D3. The analog signal corresponding to the forward scattered light is called the "forward scattered light signal," the analog signal corresponding to the side scattered light is called the "side scattered light signal," and the analog signal corresponding to the side fluorescence is called the "fluorescence signal." One pulse of each analog signal corresponds to one cell.

[0039] The analog signal is input to the A / D converter and converted to a digital signal. Figure 2(b) schematically shows the conversion to a digital signal by the A / D converter. For the sake of simplicity, this diagram shows the analog signal being directly input to the A / D converter. The level of the analog signal can be converted directly to a digital signal, but if necessary, processing such as noise reduction, baseline correction, and normalization may be performed. As shown in Figure 2(b), the A / D converter samples the forward scattered light signal, the side scattered light signal, and the fluorescence signal, starting from the point when the level of the forward scattered light signal reaches a predetermined threshold level among the analog signals input from the light receiving units D1 to D3. The A / D converter samples each analog signal at a predetermined sampling rate (for example, sampling 1024 points at 10 nanosecond intervals, sampling 128 points at 80 nanosecond intervals, or sampling 64 points at 160 nanosecond intervals).

[0040] Figure 2(c) schematically shows the waveform data obtained by sampling. Through sampling, matrix data is obtained, with elements being digital values ​​representing the analog signal levels at multiple time points, as waveform data corresponding to a single cell. In this way, the A / D conversion unit generates digital signals for forward scattered light, side scattered light, and side fluorescence corresponding to a single cell. The A / D conversion is repeated until the number of digitized cells reaches a predetermined number, or until a predetermined time has elapsed since the sample began flowing into the flow cell. As a result, a digital signal is obtained that combines the waveform data of N cells contained in a single sample, as shown in Figure 2(c). The set of sampling data for each cell (in the example in Figure 2, a set of 1024 digital values ​​at 10 nanosecond intervals from t=0ns to t=10240ns) is called waveform data, and the set of waveform data obtained from a single sample is called a digital signal.

[0041] Each waveform data generated by the A / D converter may be assigned an index to identify the respective cell. For example, the index may be assigned an integer from 1 to N in the order of the generated waveform data, and the same index may be assigned to the waveform data of forward scattered light, side scattered light, and side fluorescence obtained from the same cell.

[0042] Since each waveform data corresponds to one cell, the index corresponds to the cell that was measured. By assigning the same index to waveform data corresponding to the same cell, the deep learning algorithm described later can analyze the waveform data of forward scattered light, side scattered light, and fluorescence corresponding to each cell as a set and classify the cell type.

[0043] <Generating training data> Figure 3 is a schematic diagram illustrating an example of a method for generating training data used to train a deep learning algorithm for determining cell types. The training data 75 is waveform data generated based on the analog signals of forward scattered light (FSC) 70a, side scattered light (SSC) 70b, and side fluorescence (SFL) 70c obtained from cells contained in a sample measured by a flow cytometer. The method for acquiring the waveform data is as described above.

[0044] Training data 75 can be obtained, for example, by measuring a sample using a flow cytometer and analyzing the cells contained in the sample based on a conventional scattergram, using waveform data of cells that are judged to be highly likely to be of a specific cell type. To illustrate with an example using a hemocytometer, first, a blood sample is measured with a flow cytometer, and waveform data of forward scattered light, side scattered light, and fluorescence of individual cells contained in the sample are accumulated. Based on the side scattered light intensity (pulse height of the side scattered light signal) and fluorescence intensity (pulse height of the fluorescence signal), the cells are classified into groups of neutrophils, lymphocytes, monocytes, eosinophils, basophils, immature granulocytes, and abnormal cells. Training data can be obtained by assigning label values ​​corresponding to the classified cell type to the waveform data of those cells. For example, the mode, mean, or median of the side scattered light intensity and side fluorescence intensity of cells contained in the neutrophil group can be determined, representative cells can be identified based on these values, and training data can be obtained by assigning the label value "1" corresponding to neutrophils to the waveform data of those cells. The method for generating training data is not limited to this; for example, training data may be obtained by collecting only specific cells using a cell sorter, measuring those cells with a flow cytometer, and assigning cell labels to the resulting waveform data.

[0045] The analog signals 70a, 70b, and 70c represent the forward scattered light signal, the side scattered light signal, and the side fluorescence signal, respectively, when neutrophils are measured by a flow cytometer. When these analog signals are A / D converted as described above, waveform data 72a for the forward scattered light signal, waveform data 72b for the side scattered light signal, and waveform data 72c for the side fluorescence signal are obtained. Within each of the waveform data 72a, 72b, and 72c, adjacent cells store the signal level at intervals corresponding to the sampling rate, for example, 10 nanosecond intervals. The waveform data 72a, 72b, and 72c are combined with label values ​​77 that represent the type of cell from which the data originated, and input into the deep learning algorithm 50 as training data 75 so that each cell has a set of three waveform data, in other words, three signal intensity data (signal intensity of forward scattered light, signal intensity of side scattered light, and signal intensity of side fluorescence). In the example in Figure 3, since the cells from which the training data originated are neutrophils, the waveform data 72a, 72b, and 72c are assigned the label value 77 "1" to indicate that they are neutrophils, and training data 75 is generated. Figure 4 shows an example of label value 77. Since training data 75 is generated for each cell type, different label values ​​77 are assigned depending on the cell type.

[0046] <Overview of Deep Learning> Figure 3 is used as an example to illustrate the overview of neural network training. The neural network 50 is preferably a convolutional neural network having convolutional layers. The number of nodes in the input layer 50a of the neural network 50 corresponds to the number of elements in the arrays contained in the waveform data of the input training data 75. The number of elements in the arrays is equal to the sum of the number of elements in the waveform data 72a, 72b, and 72c for forward scattered light, side scattered light, and side fluorescence, corresponding to one cell. In the example in Figure 3, each of the waveform data 72a, 72b, and 72c contains 1024 elements, so the number of nodes in the input layer 50a is 1024 × 3 = 3072. The waveform data 72a, 72b, and 72c are input to the input layer 50a of the neural network 50. The label values ​​77 of each waveform data in the training data 75 are input to the output layer 50b of the neural network to train the neural network 50. The code 50c in Figure 3 indicates the hidden layer.

[0047] <Method for analyzing waveform data> Figure 5 shows an example of a method for analyzing waveform data of cells to be analyzed. In this waveform data analysis method, analysis data 85 is generated from the forward scattered light analog signal 80a, the side scattered light analog signal 80b, and the side fluorescence analog signal 80c obtained from the cells to be analyzed by a flow cytometer, and consists of waveform data obtained by the method described above.

[0048] It is preferable that the analysis data 85 and the training data 75 have at least the same acquisition conditions. The acquisition conditions include the conditions for measuring the cells contained in the sample using a flow cytometer, such as the preparation conditions of the measurement sample, the flow rate when the measurement sample is flowed through the flow cell, the intensity of the light irradiated onto the flow cell, and the amplification factor of the light-receiving unit that receives scattered light and fluorescence. The acquisition conditions also include the sampling rate when performing A / D conversion of the analog signal.

[0049] As the cells to be analyzed flow through the flow cell, an analog signal 80a of forward scattered light, an analog signal 80b of side scattered light, and an analog signal 80c of side fluorescence are obtained. When these analog signals 80a, 80b, and 80c are converted using A / D as described above, the timing of signal intensity acquisition is synchronized for each cell, resulting in waveform data 82a of the forward scattered light signal, waveform data 82b of the side scattered light signal, and waveform data 82c of the side fluorescence signal. The waveform data 82a, 82b, and 82c are combined into sets of data representing the three signal intensities of each cell (signal intensity of forward scattered light, signal intensity of side scattered light, and signal intensity of side fluorescence), and input into the deep learning algorithm 60 as analysis data 85.

[0050] When the analysis data 85 is input to the input layer 60a of the neural network 60 that constitutes the trained deep learning algorithm 60, the output layer 60b outputs the analysis result 83 as classification information regarding the cell type corresponding to the analysis data 85. In Figure 5, the symbol 60c indicates the hidden layer. The classification information regarding cell type is, for example, the probability that a cell belongs to each of several cell types. Furthermore, it is determined that the cell being analyzed from which the analysis data 85 was obtained belongs to the classification with the highest value among these probabilities, and the analysis result 83 may include a label value 82, which is an identifier representing that cell type. The analysis result 83 may be the label value itself, or it may be data in which the label value has been replaced with information indicating the cell type (for example, a string). In Figure 5, based on the analysis data 85, the deep learning algorithm 60 outputs the label value "1", which had the highest probability of belonging to the cell being analyzed from which the analysis data 85 was obtained, and further outputs the text data "neutrophil" corresponding to this label value as the analysis result 83. The deep learning algorithm 60 may output the label values, but another computer program may output the most preferable label values ​​based on the probabilities calculated by the deep learning algorithm 60.

[0051] [2. Cell analysis devices and measurement of biological samples using cell analysis devices] The waveform data of cells in this embodiment, or the analog signal of the cells from which it is derived, can be acquired by a first cell analyzer 4000 or a second cell analyzer 4000'. Figure 6(a) shows an example of the appearance of the cell analyzer 4000. Figure 6(b) shows an example of the appearance of the cell analyzer 4000'. In Figure 6(a), the cell analyzer 4000 includes a measurement unit 400 and a processing unit 300 for setting the measurement conditions and controlling the measurement of a sample in the measurement unit 400, and for analyzing the results of the cell data analysis by the deep learning algorithm 60. In Figure 6(b), the cell analyzer 4000' includes a measurement unit 500 and a processing unit 300 for setting the measurement conditions and controlling the measurement of a sample in the measurement unit 500, and for analyzing the results of the cell data analysis by the deep learning algorithm 60. The measurement units 400, 500 and the processing unit 300 can be connected to each other by wired or wireless means so as to be able to communicate with each other. The following shows examples of the configurations of the measurement units 400 and 500, but this embodiment is not limited to the following examples. The processing unit 300 may be shared with the vendor-side device 100, which will be described later.

[0052] <Preparation of the first cell analyzer and sample> (Configuration of the measurement unit and processing unit) This section describes an example configuration where the measurement unit 400 is a blood analyzer, more specifically a blood cell counter, equipped with an FCM detection unit which is a flow cytometer for detecting cells in a blood sample.

[0053] (Configuration Example 1) Referring to Figures 7 to 19, an example configuration of the measurement unit 400 and the processing unit 300 will be described. Figure 7 shows an example block diagram of the measurement unit 400. As shown in Figure 7, the measurement unit 400 includes an FCM detection unit 410 for detecting blood cells, an analog processing unit 420 for processing the analog signal output from the FCM detection unit 410, a measurement unit control unit 480, a sample preparation unit 440, a device mechanism unit 430, and a sample aspiration unit 450.

[0054] Figure 8 is a schematic diagram illustrating the sample aspiration unit 450 and the sample preparation unit 440. The sample aspiration unit 450 includes a nozzle 451 for aspirating a blood sample (whole blood) from a blood collection tube T, and a pump 452 for applying negative / positive pressure to the nozzle. 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 with the nozzle 451 inserted into the blood collection tube T, the blood sample is aspirated through the nozzle 451. The device mechanism unit 430 may also include a hand member for inverting and agitating the blood collection tube T before aspirating blood from the blood collection tube T.

[0055] The sample preparation unit 440 is equipped with five reaction chambers 440a to 440e. These reaction chambers 440a to 440e are used for the DIFF, RET, WPC, PLT-F, and WNR measurement channels, respectively. Each reaction chamber is connected via a flow path to a hemolytic agent container containing the hemolytic agent and a staining solution container containing the staining solution, which are reagents corresponding to each measurement channel. A measurement channel is comprised of one reaction chamber and the reagents (hemolytic agent and staining solution) connected to it. For example, the DIFF measurement channel is comprised of the DIFF hemolytic agent and DIFF staining solution, which are reagents for DIFF measurement, and the DIFF reaction chamber 440a. Other measurement channels are configured similarly. While this example illustrates a configuration where one measurement channel is equipped with one hemolytic agent and one staining solution, a single measurement channel does not necessarily need to be equipped with both, and one reagent may be shared among multiple measurement channels.

[0056] The nozzle 451, which has aspirated the blood sample, is moved horizontally and vertically by the device mechanism unit 430 to access from above the reaction chamber 440a to 440e corresponding to the measurement channel corresponding to the order, and discharges the aspirated blood sample. The sample preparation unit 440 supplies the corresponding hemolytic agent and staining solution to the reaction chamber from which the blood sample was discharged, and prepares the 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 to the FCM detection unit 410 via the flow path, and cell measurement is performed by flow cytometry.

[0057] Figure 9 shows an example of the optical system configuration of the FCM detection unit 410. As shown in Figure 9, in flow cytometry, when cells contained in the sample to be measured pass through the flow cell (sheath flow cell) 4113 provided inside the flow cytometer, the light source 4111 irradiates the flow cell 4113 with light, and the scattered light and fluorescence emitted from the cells inside the flow cell 4113 are detected by this light.

[0058] In Figure 9, the light emitted from the laser diode, which is the light source 4111, is irradiated onto cells passing through the flow cell 4113 via the irradiation lens system 4112.

[0059] In this embodiment, the light source 4111 of the flow cytometer is not particularly limited, and a light source 4111 with a wavelength suitable for exciting fluorescent dyes is selected. Examples of such light sources 4111 include semiconductor laser light sources including red semiconductor laser light sources and / or blue semiconductor laser light sources, argon laser light sources, gas laser light sources such as helium-neon lasers, and mercury arc lamps. Semiconductor laser light sources are particularly preferred because they are much cheaper than gas laser light sources.

[0060] As shown in Figure 9, forward scattered light emitted from particles passing through the flow cell 4113 is received by a forward scattered light receiving element 4116 via a focusing lens 4114 and a pinhole section 4115. The forward scattered light receiving element 4116 is a photodiode. Side scattered light is received by a side scattered light receiving element 4121 via a focusing lens 4117, a dichroic mirror 4118, a bandpass filter 4119, and a pinhole section 4120. The side scattered light receiving element 4121 is a photodiode. Side fluorescence is received by a side fluorescence receiving element 4122 via a focusing lens 4117 and a dichroic mirror 4118. The side fluorescence receiving element 4122 is an avalanche photodiode. Alternatively, photomultiplier tubes may be used as the forward-scattered light receiving element 4116, the side-scattered light receiving element 4121, and the side-fluorescent light receiving element 4122.

[0061] The received signals output from each of the photodetectors 4116, 4121, and 4122 are input to the analog processing unit 420 via amplifiers 4151, 4152, and 4153, respectively.

[0062] Returning to Figure 7, the analog processing unit 420 performs noise reduction, smoothing, and other processing on the analog signal input from the FCM detection unit 410, and outputs the processed analog signal to the measurement unit control unit 480.

[0063] The measurement unit control unit 480 comprises an A / D conversion unit 482, a processor 4831, a RAM 4834, a storage unit 4835, a bus controller 4850, a parallel processing processor 4833, and an interface unit 489 that connects to the processing unit 300. Furthermore, the measurement unit control unit 480 comprises an interface unit 484 interposed between it and the A / D conversion unit 482, and an interface unit 488 interposed between it and the device mechanism unit 430. As shown in Figure 7, in this configuration example, the parallel processing processor 4833 is mounted in the cell analyzer 4000 by being incorporated inside the measurement unit 400.

[0064] The processor 4831 is connected to the interface unit 489, interface unit 488, interface unit 484, RAM 4834, and storage unit 4835 via the bus 485. The processor 4831 is also connected to the parallel processing processor 4833 via the bus 485. The processing unit 300 is connected to the various parts of the measurement unit 400 via the interface unit 489 and the bus 485. The bus 485 is, for example, a transmission line having a data transfer rate of several hundred MB / s or more. The bus 485 may also be configured as a transmission line having a data transfer rate of, for example, 1 GB / s or more. The bus 485 performs data transfer based on the PCI-Express or PCI-X standard, for example.

[0065] The A / D conversion unit 482 converts the analog signal output from the analog processing unit 420 into a digital signal. The A / D conversion unit 482 converts the analog signal from the start to the end of the measurement of the sample into a digital signal. If multiple types of analog signals (for example, analog signals corresponding to forward scattered light intensity, side scattered light intensity, and fluorescence intensity, respectively) are generated by measurement on a certain measurement channel, the A / D conversion unit 482 converts each analog signal from the start to the end of the measurement into a digital signal. 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, as explained with reference to Figure 9, for example. The A / D conversion unit 482 converts each of the analog signals input from the multiple signal transmission paths 421 into a digital signal. Each signal transmission path 421 is configured, for example, to transmit the analog signal as a differential signal.

[0066] As explained with reference to Figure 2, the A / D converter 482 samples the analog signal at a predetermined sampling rate (for example, sampling 1024 points at 10 nanosecond intervals, sampling 128 points at 80 nanosecond intervals, or sampling 64 points at 160 nanosecond intervals). The A / D converter 482 performs sampling on three types of analog signals corresponding to each cell, thereby generating waveform data for the forward scattered light signal, the side scattered light signal, and the fluorescence signal for each cell. The A / D converter 482 assigns an index to each of the generated waveform data. As shown in Figure 2(c), the generated waveform data becomes a digital signal created by successively generating the waveform data of N cells contained in a single sample. This generates three digital signals corresponding to the three types of analog signals (forward scattered light signal, side scattered light signal, and fluorescence signal) obtained from the N cells.

[0067] In addition to generating waveform data from the analog signal, the A / D conversion unit 482 may also calculate peak values ​​from the pulses of the analog signal.

[0068] The A / D converter 482 inputs the generated digital signal to the bus 485. The bus controller 4850 transmits the digital signal output from the A / D converter 482 to the RAM 4834, for example, by DMA (Direct Memory Access) transfer. The RAM 4834 stores the digital signal.

[0069] The processor 4831 uses a parallel processing processor 4833 to perform analysis processing of waveform data contained in the generated digital signal according to the deep learning algorithm 60. That is, the processor 4831 is programmed to perform analysis processing of waveform data contained in the digital signal according to the deep learning algorithm 60. Analysis software 4832 for analyzing cell data based on the deep learning algorithm 60 may be stored in the memory unit 4835. In this case, the processor 4831 performs data analysis processing based on the deep learning algorithm 60 by executing the analysis software 4832 stored in the memory unit 4835. The processor 4831 is, for example, a CPU (Central Processing Unit). The processor 4831 may be, for example, an Intel Core i9, Core i7, Core i5, or an AMD Ryzen 9, Ryzen 7, Ryzen 5, Ryzen 3, etc.

[0070] Processor 4831 controls the parallel processing processor 4833. The parallel processing processor 4833 performs parallel processing, such as matrix operations, in response to control by processor 4831. In other words, processor 4831 is the master processor of the parallel processing processor 4833, and the parallel processing processor 4833 is the slave processor of processor 4831. Processor 4831 is also called the host processor or main processor.

[0071] The parallel processing processor 4833 executes multiple arithmetic operations in parallel, which are at least part of the processing related to waveform data analysis. The parallel processing processor 4833 is, for example, a GPU (Graphics Processing Unit), an FPGA (Field Programmable Gate Array), or an ASIC (Application Specific Integrated Circuit). If the parallel processing processor 4833 is an FPGA, it may have, for example, pre-programmed arithmetic operations related to a trained deep learning algorithm 60. If the parallel processing processor 4833 is an ASIC, it may have, for example, pre-built circuits for executing arithmetic operations related to a trained deep learning algorithm 60, or it may have programmable modules built in addition to such built-in circuits. The parallel processing processor 4833 may use, for example, NVIDIA's GeForce, Quadro, TITAN, Jetson, etc. If using the Jetson series, for example, Jetson Nano, Jetson Tx2, Jetson Xavier, or Jetson AGX Xavier can be used.

[0072] The processor 4831 performs calculations related to the control of the measurement unit 400, for example. The processor 4831 performs calculations related to the control signals transmitted and received between the device mechanism unit 430, the sample preparation unit 440, and the sample aspiration unit 450, for example. The processor 4831 performs calculations related to the transmission and reception of information with the processing unit 300, for example. The processor 4831 performs processes related to reading program data from the storage unit 4835, expanding the program into the RAM 4834, and transmitting and receiving data with the RAM 4834, for example. Each of the above processes performed by the processor 4831 is required to be executed in a predetermined order, for example. For example, if the processes required to control the device mechanism unit 430, the sample preparation unit 440, and the sample aspiration unit 450 are A, B, and C, they may be required to be executed in the order B, A, C. Because the processor 4831 often performs sequential processes that depend on such an order, increasing the number of arithmetic units (sometimes called "processor cores," "cores," etc.) does not necessarily increase the processing speed.

[0073] On the other hand, the parallel processing processor 4833 performs routine and large-scale computational processing, such as operations on matrix data containing a large number of elements. In this embodiment, the parallel processing processor 4833 performs parallel processing that parallelizes at least a portion of the process of analyzing waveform data according to the deep learning algorithm 60. The deep learning algorithm 60 includes, for example, a large number of matrix operations. The deep learning algorithm 60 may include, for example, at least 100 matrix operations, and may also include at least 1000 matrix operations. The parallel processing processor 4833 has multiple arithmetic units, each of which can perform matrix operations simultaneously. In other words, the parallel processing processor 4833 can perform matrix operations by each of the multiple arithmetic units in parallel as parallel processing. For example, the matrix operations included in the deep learning algorithm 60 can be divided into multiple operations that are not order-dependent. These divided operations can be executed in parallel by each of the multiple arithmetic units. These arithmetic units are sometimes called "processor cores," "cores," etc.

[0074] By performing such parallel processing, it is possible to speed up the overall computational processing of the measurement unit 400. Processing such as matrix operations included in the deep learning algorithm 60 is sometimes called "Single Instruction Multiple Data" (SIMD). The parallel processing processor 4833 is suitable for such SIMD operations. Such a parallel processing processor 4833 is sometimes called a vector processor.

[0075] As described above, processor 4831 is suitable for performing diverse and complex processing. On the other hand, parallel processing processor 4833 is suitable for performing a large amount of standardized processing in parallel. By performing a large amount of standardized processing in parallel, the Turn Around Time (TAT) required for computation is shortened.

[0076] Furthermore, the parallel processing performed by the parallel processing processor 4833 is not limited to matrix operations. For example, when the parallel processing processor 4833 performs learning operations according to the deep learning algorithm 50, differential operations and other operations related to the learning process may also be subject to parallel processing.

[0077] The number of arithmetic units in processor 4831 can be, for example, dual-core (2 cores), quad-core (4 cores), or octa-core (8 cores). On the other hand, parallel processing processor 4833 has, for example, at least 10 arithmetic units (10 cores) and can perform 10 matrix operations in parallel. Some parallel processing processors 4833 have, for example, several tens of arithmetic units. Also, some parallel processing processors 4833 have, for example, at least 100 arithmetic units (100 cores) and can perform 100 matrix operations in parallel. Some parallel processing processors 4833 have, for example, several hundred arithmetic units. Also, some parallel processing processors 4833 have, for example, at least 1000 arithmetic units (1000 cores) and can perform 1000 matrix operations in parallel. Some parallel processing processors, such as the 4833, have thousands of arithmetic units.

[0078] Figure 11 shows an example configuration of a parallel processing processor 4833. The parallel processing processor 4833 includes multiple arithmetic units 4836 and RAM 4837. Each of the arithmetic units 4836 performs matrix data arithmetic operations in parallel. RAM 4837 stores data related to the arithmetic operations performed by the arithmetic units 4836. RAM 4837 is memory with a capacity of at least 1 gigabyte. RAM 4837 may also be memory with a capacity of 2 gigabytes, 4 gigabytes, 6 gigabytes, 8 gigabytes, or 10 gigabytes or more. The arithmetic units 4836 retrieve data from RAM 4837 and perform arithmetic operations. The arithmetic units 4836 are sometimes called "processor cores," "cores," etc.

[0079] Figures 12 to 14 show examples of mounting the parallel processing processor 4833 on the measurement unit 400. Figures 12 to 14 show an example in which the parallel processing processor 4833 is mounted on the cell analyzer 4000 in a manner that it is integrated inside the measurement unit 400. Figures 12 and 13 show an example of mounting in which the processor 4831 and the parallel processing processor 4833 are provided as separate components. As shown in Figure 12, the processor 4831 is mounted on, for example, a circuit board 4838. The parallel processing processor 4833 is mounted on, for example, a graphics board 4830, and the graphics board 4830 is connected to the circuit board 4838 via a connector 4839. The processor 4831 is connected to the parallel processing processor 4833 via a bus 485. As shown in Figure 13, the parallel processing processor 4833 may be mounted directly on, for example, the circuit board 4838 and connected to the processor 4831 via a bus 485. Figure 14 shows an example of a configuration in which the processor 4831 and the parallel processing processor 4833 are provided as an integrated unit. As shown in Figure 14, the parallel processing processor 4833 may be built into the processor 4831 mounted on the substrate 4838, for example.

[0080] Figure 15 shows another example of mounting the parallel processing processor 4833 on the measurement unit 400. Figure 15 shows an example of mounting the parallel processing processor 4833 on the measurement unit 400 by an external device 4800 connected to the measurement unit 400. The parallel processing processor 4833 is implemented in the external device 4800, which is, for example, a USB (Universal Serial Bus) device, and this USB device is connected to the bus 485 via the interface unit 487, thereby mounting the parallel processing processor 4833 on the cell analysis device 4000. The USB device may be a small device such as a USB dongle. The interface unit 487 is, for example, a USB interface with a transfer speed of several hundred Mbps, and more preferably a USB interface with a transfer speed of several Gbps to several tens of Gbps or more. As the external device 4800 on which the parallel processing processor 4833 is implemented, for example, an Intel Neural Compute Stick 2 may be used.

[0081] Multiple parallel processing processors 4833 may be installed in the cell analysis device 4000 by connecting multiple USB devices, each equipped with a parallel processing processor 4833, to the interface unit 487. Since the number of arithmetic units 4836 in a single USB device may be smaller than that of a GPU, etc., the number of cores can be scaled up by adding multiple USB devices connected to the measurement unit 400.

[0082] Figures 16, 17, and 18 show an overview of the arithmetic processing performed by the parallel processing processor 4833 based on the control of the analysis software 4832 running on the processor 4831. Figure 16 shows an example configuration of the parallel processing processor 4833 that performs the arithmetic processing. The parallel processing processor 4833 has multiple arithmetic units 4836 and RAM 4837. The processor 4831, which runs the analysis software 4832, can instruct the parallel processing processor 4833 to perform at least some of the arithmetic processing required when analyzing waveform data with the deep learning algorithm 60. The processor 4831 instructs the parallel processing processor 4833 to perform arithmetic processing related to waveform data analysis based on the deep learning algorithm. All or at least part of the waveform data corresponding to the signal detected by the FCM detection unit 410 is stored in RAM 4834. The data stored in RAM 4834 is transferred to RAM 4837 of the parallel processing processor 4833. Data stored in RAM4834 is transferred to RAM4837, for example, using the DMA (Direct Memory Access) method. Each of the multiple arithmetic units 4836 of the parallel processing processor 4833 performs arithmetic operations on the data stored in RAM4837 in parallel. Each of the multiple arithmetic units 4836 retrieves the necessary data from RAM4837 and performs the arithmetic operation. Data corresponding to the calculation result is stored in RAM4837 of the parallel processing processor 4833. Data corresponding to the calculation result is transferred from RAM4837 to RAM4834, for example, using the DMA method.

[0083] Figure 17 shows an overview of matrix operations performed by the parallel processing processor 4833. When analyzing waveform data according to the deep learning algorithm 60, matrix multiplication (matrix operations) is performed. The parallel processing processor 4833 executes multiple matrix operations in parallel, for example. Figure 17(a) shows the formula for matrix multiplication. In the formula shown in (a), matrix c is obtained by multiplying an n x n matrix a and an n x n matrix b. As illustrated in Figure 17, the formula is written using a multi-level loop construct. Figure 17(b) shows an example of operations executed in parallel by the parallel processing processor 4833. The formula illustrated in Figure 17(a) can be divided into n x n operations, for example, which is the number of combinations of the first-level loop variable i and the second-level loop variable j. Since each of these divided operations is independent of the others, they can be executed in parallel.

[0084] Figure 18 is a conceptual diagram illustrating how the multiple arithmetic operations exemplified in Figure 17(b) are executed in parallel by the parallel processing processor 4833. As shown in Figure 18, each of the multiple arithmetic operations is assigned to one of the multiple arithmetic units 4836 provided by the parallel processing processor 4833. Each of the arithmetic units 4836 executes its assigned arithmetic operations in parallel with the others. In other words, each of the arithmetic units 4836 executes the divided arithmetic operations simultaneously.

[0085] As illustrated in Figures 17 and 18, the parallel processing processor 4833 performs calculations to obtain, for example, information regarding the probability that the cells corresponding to the waveform data belong to each of several cell types. Based on the calculation results, the processor 4831, which executes the analysis software 4832, performs an analysis on the cell type of the cells corresponding to the waveform data. The calculation results are stored in the RAM 4837 of the parallel processing processor 4833 and transferred from RAM 4837 to RAM 4834. The processor 4831 transmits the results of the analysis based on the calculation results stored in RAM 4834 to the processing unit 300 via the bus 485 and interface unit 489.

[0086] The calculation of the probability that a cell belongs to each of the multiple cell types may be performed by a processor other than the parallel processing processor 4833. For example, the calculation result from the parallel processing processor 4833 may be transferred from RAM 4837 to RAM 4834, and the processor 4831 may calculate information regarding the probability that the cell corresponding to each waveform data belongs to each of the multiple cell types based on the calculation result read from RAM 4834. Alternatively, the calculation result from the parallel processing processor 4833 may be transferred from RAM 4837 to the processing unit 300, and a processor mounted on the processing unit 300 may calculate information regarding the probability that the cell corresponding to each waveform data belongs to each of the multiple cell types.

[0087] In this embodiment, the processes shown in Figures 17 and 18 are applied, for example, to the computational processing (also called filtering) related to the convolutional layer in the deep learning algorithm 60.

[0088] Figure 19 shows an overview of the computational processing related to the convolutional layer. Figure 19(a) shows an example of forward-scattered light (FSC) waveform data as the waveform data input to the deep learning algorithm 60. The waveform data in this embodiment is one-dimensional matrix data as shown in Figure 2. More simply put, the waveform data is an array in which the elements are arranged in a single column. Here, for the sake of explanation, the number of elements in the waveform data is n (where n is an integer of 1 or more). Multiple filters are shown in (a). The filters are generated by the learning process of the deep learning algorithm 50. Each of the multiple filters is one-dimensional matrix data representing the features of the waveform data. The filter shown in (a) is a 1x3 matrix data, but the number of columns is not limited to 3. By performing matrix operations on the waveform data input to the deep learning algorithm 60 and each filter, features corresponding to the cell type related to the waveform data are calculated. Figure 19(b) shows an overview of the matrix operations between the waveform data and the filters. As shown in (b), the matrix operation is performed by shifting each filter by one position for each element of the waveform data. The matrix operation is calculated using Equation 1 below.

number

[0089] The parallel processing processor 4833 executes the matrix operation represented by Equation 1 in parallel using each of the multiple processing units 4836. Based on the calculations performed by the parallel processing processor 4833, classification information regarding the cell type of each cell is generated. The generated classification information is transmitted to the processing unit 300, which is used to generate and display the test results of the sample based on the classification information.

[0090] The measurement unit 400 can process waveform data in association with identification information. Specifically, the measurement unit 400 can generate waveform data analysis results (i.e., classification information regarding the cell type of each cell) in association with identification information. For example, the measurement unit 400 associates the classification information regarding the cell type of each cell with the identification information and transmits it to the processing unit 300. Identification information includes, for example, (1) identification information of the biological sample corresponding to the waveform data, (2) identification information of the cell corresponding to the waveform data (e.g., the index mentioned above), (3) identification information of the patient corresponding to the waveform data, (4) identification information of the test corresponding to the waveform data, (5) identification information of the cell analyzer on which the waveform data was measured, and (6) identification information of the facility such as a hospital or testing facility (hereinafter referred to as "test-related facility") on which the waveform data was measured. (1) The identification information of the biological sample corresponding to the waveform data may include information for determining the priority of parallel processing, such as information regarding the time when a measurement order for the biological sample was registered, information regarding the time when the analyzer identified the biological sample, information regarding the time when the analyzer started measuring the biological sample, information for identifying whether the biological sample is an emergency sample or a normal sample, and information for identifying whether the biological sample is a re-measurement or a new measurement. The measurement unit 400 can, for example, obtain at least one of the above identification information (1) to (6) or a combination thereof from the LIS (Laboratory Information System) or the processing unit 300 when it receives a test order from the LIS or the processing unit 300. For example, at least one of the exemplified (1) to (6) is transmitted to the processing unit 300 in association with classification information and provided to the user as a test result via the processing unit 300. Multiple combinations of the exemplified (1) to (6) may be transmitted to the processing unit 300 in association with classification information. The measurement unit 400 can also, for example, generate at least one of the above identification information (1) to (6) or a combination thereof itself.

[0091] The association between waveform data and the above-mentioned identification information (at least one of (1) to (6) above, or a combination thereof) may be performed by the processing unit 300. For example, when the processing unit 300 receives a test order from the LIS, it acquires identification information for the biological sample and identification information for the patient. The processing unit 300 instructs the measurement unit 400 to perform the measurement corresponding to the above-mentioned test order. When the processing unit 300 acquires the result (i.e., waveform data) corresponding to the measurement instruction from the measurement unit 400, it associates the result (i.e., waveform data) with the above-mentioned identification information.

[0092] Returning to Figure 6, the processing unit 300 is connected to the processor 4831 via the interface unit 489 and the bus 485, and can receive analysis results from the processor 4831 and the parallel processing processor 4833. The interface unit 489 is, for example, a USB interface.

[0093] Figure 10 shows the configuration of the processing unit 300. The processing unit 300 comprises a processor 3001, a bus 3003, a storage unit 3004, an interface unit 3006, a display unit 3015, and an operation unit 3016. The processing unit 300 is composed of a general-purpose personal computer as hardware and functions as the processing unit of the cell analyzer 4000 by executing a dedicated program stored in the storage unit 3004.

[0094] The processor 3001 is a CPU and is capable of executing programs stored in the memory unit 3004.

[0095] The storage unit 3004 includes a hard disk drive. The storage unit 3004 stores at least a program 60 for processing cell classification information transmitted from the measurement unit 400 and generating the test results of the sample. The test results of the sample refer to the result of counting the blood cells contained in the sample based on the analysis results 83 of the classification information 82 of individual cells obtained by the measurement unit 400, as described later.

[0096] The display unit 3015 is equipped with 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 receives an image signal input from the processor 3001 and can display the analysis result 83 received from the measurement unit 400 and the inspection result obtained by the processor 3001 analyzing the analysis result 83.

[0097] The control unit 3016 is equipped with a pointing device including a keyboard, mouse, or touch panel. Users such as physicians and laboratory technicians can input measurement orders into the cell analyzer 4000 and input measurement instructions according to the measurement orders by operating the control unit 3016. The control unit 3016 can also receive instructions from the user to display the test results. Users can operate the control unit 3016 to view various information related to the test results, such as graphs, charts, and flag information assigned to samples.

[0098] The measurement unit 400 described above is connected to the processing unit 300 via the interface unit 3006. The processor 4831 of the measurement unit 400 can transmit the classification information of individual cells generated by the deep learning algorithm 60, associated with the identification information of the sample, to the processor 3001 of the processing unit 300. The processor 3001 stores the cell analysis results 83 received from the measurement unit 400 in the storage unit 3004, associated with the identification information of the sample.

[0099] <Operation of the cell analyzer> Refer to Figures 20 to 22 to explain the analysis operation of the cell analyzer 4000.

[0100] When the processor 3001 of the processing unit 300 receives a measurement order and measurement instruction from the user via the operation unit 3016, it sends a measurement command to the measurement unit 400 (step S1).

[0101] When the processor 4831 of the measurement unit 400 receives a measurement command, it starts measuring the sample. The processor 4831 instructs the sample aspiration unit 450 to aspirate the sample from the blood collection tube T (step S10). Next, the processor 4831 instructs the sample aspiration unit 450 to dispense the aspirated sample into one of the reaction chambers 440a to 440e of the sample preparation unit 440. The measurement command transmitted from the processing unit 300 in step S1 includes information on the measurement channel for which measurement is requested by the measurement order. Based on the measurement channel information included in the measurement command, the processor 4831 controls the sample aspiration unit 450 to dispense the sample into the reaction chamber of the corresponding measurement channel.

[0102] The processor 4831 instructs the sample preparation unit 440 to prepare the measurement sample (step S11). Specifically, the sample preparation unit 440, upon receiving a command from the processor 4831, supplies reagents (hemolytic agent and staining solution) to the reaction chamber from which the sample was discharged, and mixes the sample and reagents. This prepares a measurement sample in which red blood cells are lysed 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 solution.

[0103] The processor 4831 instructs the FCM detection unit 410 to measure the prepared sample (step S12). Specifically, the processor 4831 controls the device mechanism unit 430 to deliver the sample from the reaction chamber of the sample preparation unit 440 to the FCM detection unit 410. The reaction chamber and the FCM detection unit 410 are connected by a flow channel, and the sample delivered from the reaction chamber flows through the flow cell 4113 and is irradiated with laser light by the light source 4111 (see Figure 9). When the cells contained in the sample pass through the flow cell 4113, light is irradiated onto the cells, and the forward scattered light, side scattered light, and side fluorescence generated from the cells are detected by the photodetectors 4116, 4121, and 4122, respectively, and an analog signal corresponding to the light detection intensity is output. The analog signal is output to the A / D conversion unit 482 via the analog processing unit 420.

[0104] The A / D converter 482 generates a digital signal containing waveform data of individual cells by sampling the analog signal at a predetermined rate (step S13). The method for generating the waveform data and digital signal has already been described. The processor 4831 takes the digital signal generated by the A / D converter 482 into the RAM 4834. For example, the processor 4831 controls the bus controller 4850 to take the digital signal generated by the A / D converter 482 into the RAM 4834 via DMA transfer. By DMA transfer, the digital signal is transferred directly to the RAM 4834 without going through the processor 4831. Specifically, the processor 4831 takes the digital signals of the forward scattered light signal, the side scattered light signal, and the fluorescence signal obtained from the cells contained in the sample to be examined into the RAM 4834. The digital signals are stored in the RAM 4834.

[0105] The processor 4831 uses the deep learning algorithm 60 to perform cell classification based on the waveform data contained in the generated digital signal (step S14). The process in S14 will be described later.

[0106] The processor 4831 transmits the analysis results 83, which include classification information 82 of individual cells obtained as a result of S14, to the processing unit 300 in association with the identification information of the sample (step S15). For example, the analysis results 83 of multiple cells contained in a single sample are each linked to the identification information of the sample and sent to the processing unit 300.

[0107] When the processor 3001 of the processing unit 300 receives the analysis result 83 from the measurement unit (step S2), it analyzes the analysis result 83 using the program stored in the memory unit 3004 and generates the test result for the sample (step S3). In the processing of S3, for example, the number of cells is counted for each cell type based on the label values ​​included in the analysis result 83 of individual cells. For example, if there are N classification pieces from one sample that are assigned the label value "1" indicating neutrophils, a count result of neutrophil count = N is obtained as the test result for the sample.

[0108] The processor 3001 acquires counting results for measurement items corresponding to the measurement channel based on the analysis results 83 and stores them in the storage unit 3004 along with the sample identification information. Measurement items corresponding to the measurement channel are items for which counting results are requested by the measurement order. For example, the measurement items corresponding to the DIFF channel are the five types of leukocytes, namely monocytes, neutrophils, lymphocytes, eosinophils, and basophils. The measurement items corresponding to the RET channel are the number of reticulocytes. The measurement items corresponding to PLT-F are the number of platelets. The measurement items corresponding to WPC are the number of hematopoietic progenitor cells. The measurement items corresponding to WNR are the number of leukocytes and nucleated erythrocytes. The counting results are not limited to the items for which measurement is requested as listed above (also called reportable items), but may also include counting results for other cells that can be measured by the same measurement channel. For example, if the measurement channel is DIFF, as shown in Figure 4, in addition to the five types of leukocytes, immature granulocytes (IG) and abnormal cells are also included in the counting results. Furthermore, the processor 3001 generates the test results of the sample by analyzing the obtained counting results and stores them in the memory unit 3004. The analysis of the counting results includes determining, for example, whether the counting results are within the normal range, whether any abnormal cells have been detected, and whether the deviation from the previous test results is within an acceptable range.

[0109] The processor 3001 displays the generated test results on the display unit 3015 (step S4).

[0110] Next, with reference to Figure 21, the cell classification process in step S14 will be described. The cell classification process in step S14 is performed by the processor 4831 in accordance with the operation of the analysis software 4832. The processor 4831 transfers the digital signal acquired by RAM 4834 in step S13 to the parallel processing processor 4833 (S101). As shown in Figure 16, the processor 4831 transfers the digital signal from RAM 4834 to RAM 4837 by DMA transfer. The processor 4831 controls, for example, the bus controller 4850 to perform a DMA transfer of the digital signal from RAM 4834 to RAM 4837.

[0111] Processor 4831 instructs parallel processing processor 4833 to perform parallel processing on the waveform data contained in the digital signal (S102). Processor 4831 instructs parallel processing by calling, for example, a kernel function of parallel processing processor 4833. The processing performed by parallel processing processor 4833 will be described later in the flowchart illustrated in Figure 22. Processor 4831 instructs parallel processing processor 4833 to perform matrix operations related to the deep learning algorithm 60, for example. The digital signal is decomposed into multiple waveform data and sequentially input to the deep learning algorithm 60. The index corresponding to each cell contained in the digital signal is not input to the deep learning algorithm 60. The waveform data input to the deep learning algorithm 60 is processed by parallel processing processor 4833.

[0112] Processor 4831 receives the calculation result performed by parallel processing processor 4833 (S103). The calculation result is transferred via DMA from RAM 4837 to RAM 4834, for example, as shown in Figure 16.

[0113] The processor 4831 generates analysis results for the cell type of each measured cell based on the calculation results from the parallel processing processor 4833 (S104).

[0114] Figure 22 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 4832.

[0115] The processor 4831, which executes the analysis software 4832, causes the parallel processing processor 4833 to assign arithmetic operations to the arithmetic unit 4836 (S110). The processor 4831 causes the parallel processing processor 4833 to assign arithmetic operations to the arithmetic unit 4836, for example, by calling the kernel function of the parallel processing processor 4833. As shown in Figure 18, for example, matrix operations related to the deep learning algorithm 60 are divided into multiple arithmetic operations, and each divided arithmetic operation is assigned to the arithmetic unit 4836. Multiple waveform data are sequentially input to the deep learning algorithm 60. Matrix operations corresponding to the waveform data are divided into multiple arithmetic operations and assigned to the arithmetic unit 4836.

[0116] Each arithmetic operation is processed in parallel by multiple arithmetic units 4836 (S111). The arithmetic operations are performed on multiple waveform data.

[0117] The calculation results generated by parallel processing by multiple arithmetic units 4836 are transferred from RAM 4837 to RAM 4834 (S112). For example, the calculation results are transferred from RAM 4837 to RAM 4834 by DMA.

[0118] (Configuration example 2) Referring to Figures 23 and 24, another configuration example of the cell analyzer 4000, which consists of a measurement unit 400 and a processing unit 300, will be described. In this configuration example, the parallel processing processor is provided in the processing unit 300.

[0119] Figure 23 shows another example of a block diagram of the measurement unit 400. The measurement unit 400 shown in Figure 23 has the same configuration and functions as the measurement unit 400 described in Figures 6 and 7 and their related descriptions, except that it does not include an A / D converter 482, a processor 4831, a RAM 4834, a storage unit 4835, or a parallel processing processor 4833, and is provided with a connection port 4201. A connection cable 4202 is connected to the connection port 4201.

[0120] Figure 24 shows an example block diagram of the processing unit 300. As shown in Figure 24, the processing unit 300 comprises a processor 3001, a parallel processing processor 3002, a storage unit 3004, a RAM 3005, an interface unit 3006, an A / D conversion unit 3008, a bus controller 4850, and an interface unit 3009, all of which are connected to the bus 3003. In other words, in the example of Figure 24, the parallel processing processor 3002 is incorporated into the processing unit 300 and mounted on the cell analysis device 4000. The bus 3003 is, for example, a transmission line with a data transfer rate of several hundred MB / s or more. The bus 3003 may also be a transmission line with a data transfer rate of 1 GB / s or more. The bus 3003 performs data transfer based on standards such as PCI-Express or PCI-X. The configuration of processor 3001, parallel processing processor 3002, memory unit 3004, and RAM 3005, and the processing performed by them, are the same as the configuration and processing of processor 4831, parallel processing processor 4833, memory unit 4835, and RAM 4834 described in Figures 11 to 19 above. The A / D conversion unit 3008 samples the analog signal output from the measurement unit 400 as described above and generates a digital signal including cell waveform data. The method for generating the digital signal is as described above. In the examples of Figures 23 and 24, the connection cable 4202 has a number of transmission paths corresponding to the type of analog signal transmitted from the measurement unit 400 to the processing unit 300. For example, the connection cable 4202 is made of twisted pair cable and has a number of pairs of wires corresponding to the type of analog signal transmitted to the processing unit 300. The transmission path from connection port 3007 to A / D conversion unit 3008 may also have a number of wires corresponding to the type of analog signal transmitted to the processing unit 300. In the transmission path from connection port 3007 to A / D conversion unit 3008, for example, analog signals are transmitted as differential signals.

[0121] As shown in Figure 25, the processor 3001 and the parallel processing processor 3002 have the same configuration and functions as the processor 4831 and the parallel processing processor 4833 described above. The parallel processing processor 3002 includes a plurality of arithmetic units 3200 and RAM 3201. Analysis software 3100 for analyzing the cell type of measured cells is executed on the processor 3001. 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 this USB device may be connected to the bus 3003 via an interface unit (not shown), thereby being mounted on the cell analyzer 4000 as part of the processing unit 300. This USB device may be a small device such as a USB dongle.

[0122] As shown in Figures 23 and 24, the processing unit 300 is connected to the interface unit 489 of the measurement unit 400 via the interface unit 3006. The processing unit 300 transmits control signals from the apparatus mechanism unit 430 and the sample preparation unit 440 to the measurement unit 400 via the interface unit 3006. The interface unit 3006 is, for example, a USB interface.

[0123] In the example shown in Figures 23 and 24, the processing unit 300 is connected to the connection port 4201 of the measurement unit 400 via a connection port 3007 connected to the A / D conversion unit 3008, and a connection cable 4202 connected to the connection port 3007, in addition to the interface unit 3006. The connection port 4201 is connected to the analog processing unit 420. The analog signal output from the connection port 4201 to the processing unit 300 is, as described above, the signal processed by the analog processing unit 420 from the output of the FCM detection unit 410 of the measurement unit 400. The analog processing unit 420 performs processing, including noise reduction, 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, for example, made of a length of 1 meter or less to reduce noise during signal transmission. The analog signal is transmitted to the processing unit 300 via the connection cable 4202, for example, as a differential signal. The processing unit 300 may have multiple connection ports 3007. The processing unit 300 may acquire analog signals from multiple measurement units 400 via the multiple connection ports 3007.

[0124] The analog signal transmitted from the measurement unit 400 via the connection cable 4202 is converted into a digital signal by the A / D conversion unit 3008 of the processing unit 300. The A / D conversion unit 3008 samples the transmitted analog signal at a predetermined sampling rate (e.g., sampling 1024 points at 10 nanosecond intervals, sampling 128 points at 80 nanosecond intervals, or sampling 64 points at 160 nanosecond intervals, etc.) as described with reference to Figure 2, and generates waveform data for each cell. The waveform data is stored in the storage unit 3004 or RAM 3005 via the bus 3003. The waveform data is transferred to the RAM 3005 by DMA, for example. The processor 3001 and the parallel processing processor 3002 perform arithmetic processing on the waveform data stored in the storage unit 3004 or RAM 3005.

[0125] The analysis software 3100, which runs on the processor 3001, has the same functionality as the analysis software 4832 shown in Figure 16. By executing the analysis software 3100, the processor 3001 generates classification information regarding the cell type of the measured cells in the same manner as described in Figures 16, 17, 18, 21, 22 and related descriptions.

[0126] In the case of the cell analyzer 4000 of Configuration Example 2 shown in Figures 23 and 24, steps S13 (digital signal generation) and S14 (cell classification) of the flowchart shown in Figure 20 are performed in the processing unit 300. Step S15 (transmission of classification information) is omitted. The processing in Figures 21 and 22 is performed by the processor 3001 and the parallel processing processor 3002 of the processing unit 300.

[0127] The processor 3001 can process waveform data in association with identification information. Specifically, the processor 3001 can associate the analysis results of the waveform data (i.e., classification information regarding the cell type of each cell) with identification information. Examples of identification information include: (1) identification information of the biological sample corresponding to the waveform data, (2) identification information of the cells corresponding to the waveform data, (3) identification information of the patient corresponding to the waveform data, (4) identification information of the test corresponding to the waveform data, (5) identification information of the cell analyzer on which the waveform data was measured, and (6) identification information of the test-related facility on which the waveform data was measured. For example, when the processor 3001 receives a test order from the LIS, it can obtain at least one of the above identification information (1) to (6) or a combination thereof from the LIS. For example, at least one of the exemplified (1) to (6) is associated with classification information and provided to the user as a test result. Multiple combinations of the exemplified (1) to (6) may be associated with classification information and provided to the user as a test result.

[0128] (Configuration Example 3) Referring to Figures 26 and 27, other configuration examples of the cell analyzer 4000, which is composed of a measurement unit 400 and a processing unit 300, will be described. In this configuration example as well, the parallel processing processor 3002 is mounted in the cell analyzer 4000 by being incorporated inside the processing unit 300.

[0129] Figure 26 shows another example of a block diagram of the measurement unit 400. The measurement unit 400 shown in Figure 26 has the same configuration and functions as the measurement unit 400 described in Figures 6 and 7 and their related descriptions, except that it does not have a processor 4831, RAM 4834, storage unit 4835, or parallel processing processor 4833, and is provided with an interface unit 4851 and a transmission line 4852 for transmitting the digital signal generated by the A / D conversion unit 482 to the processing unit 300. The interface unit 4851 is, for example, an interface as a dedicated line with 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 line 4852 is, for example, a LAN cable. If the interface unit 4851 is USB 3.0, the transmission line 4852 is a USB cable compliant with USB 3.0. The transmission line 4852 is, for example, a dedicated transmission line for transmitting digital signals between the measurement unit 400 and the processing unit 300.

[0130] Figure 27 shows another example of a block diagram of the processing unit 300. The processing unit 300 shown in Figure 27 has the same configuration and function as the processing unit 300 described in Figure 24 and its related descriptions, except that it does not have an A / D conversion unit 3008 and a connection port 3007, and does have an interface unit 3010. The processing unit 300 may be connected to multiple measurement units 400 via multiple interface units 3010 and multiple interface units 3006.

[0131] Processor 3001 and parallel processing processor 3002 have the same configuration and functions as processor 3001 and parallel processing processor 3002 described in Figure 25 and related descriptions. In Figure 27, the parallel processing processor 3002 does not necessarily have to be directly connected to bus 3003; for example, it may be implemented in a USB device, and this USB device may be connected to bus 3003 via an interface unit (not shown). This USB device may be a small device such as a USB dongle.

[0132] The analysis software 3100, which runs on processor 3001, has similar functionality to the analysis software 4832 shown in Figure 16. The analysis software 3100 analyzes the cell type of the measured cells by operating in the same manner as described in Figures 16, 17, 18, 21, 22 and related descriptions.

[0133] In the case of the cell analyzer 4000 of Configuration Example 3 shown in Figures 26 and 27, step S14 (cell classification) of the flowchart shown in Figure 20 is performed in the processing unit 300. Step S15 (transmission of classification information) is omitted. The processing in Figures 21 and 22 is performed by the processor 3001 and the parallel processing processor 3002 of the processing unit 300.

[0134] In the configurations shown in Figures 26 and 27, the analog signals of cells (forward scattered light signal, side scattered light signal, side fluorescence signal) generated in the FCM detection unit 410 are converted into digital signals in the A / D conversion unit 482 within the measurement unit 400. The digital signals are sent to the processing unit 300 via the interface unit 484, bus 485, interface unit 4851, and transmission line 4852. As described above, transmission line 4852 is a dedicated transmission line 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 transmission line 4852. In other words, transmission line 4852 is a transmission line that does not involve the transmission of data related to devices other than the components constituting the cell analyzer 4000 (e.g., the measurement unit 400 and the processing unit 300). Transmission line 4852 is a separate transmission line from, for example, an intranet or the internet. This makes it possible to avoid communication speed bottlenecks in digital signal transmission even when the digital signal generated by A / D conversion is performed within the measurement unit 400 and transmitted to the processing unit 300.

[0135] (Configuration example 4) Refer to Figures 28, 29, 30, 31, and 32 to describe other configuration examples of the cell analyzer 4000.

[0136] In this configuration example, as illustrated in Figure 28, an analysis unit 600 is provided between the measurement unit 400 and the processing unit 300. In other words, in the configurations shown in Figures 28, 29, 30, 31, and 32, the cell analyzer 4000 is configured to include a measurement unit 400, a processing unit 300, and an 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 incorporated into the analysis unit 600 and mounted on the cell analyzer 4000.

[0137] The configuration of the measurement unit 400 illustrated in Figure 29 has the same configuration and functions as the measurement unit 400 described in Figure 26 and its related descriptions. 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 with 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. The transmission path 4852 is a dedicated transmission path for transmitting digital signals between the measurement unit 400 and the processing unit 300 as described above. For example, the measurement unit 400 and the processing unit 300 are connected one-to-one via the transmission line 4852.

[0138] Figure 30 shows an example configuration of the analysis unit 600. 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, all of which are connected to the bus 6003. The bus 6003 is, for example, a transmission line having a data transfer rate of several hundred MB / s or more. The bus 3003 may be a transmission line having a data transfer rate of 1 GB / s or more. The bus 3003 performs data transfer based on, for example, PCI-Express or PCI-X standards. The analysis unit 600 may be connected to multiple measurement units 400 via multiple interface units 6006. If multiple measurement units 400 are provided, each measurement unit 400 may be connected to an analysis unit 600 (for example, multiple measurement units 400 and multiple analysis units 600 are connected one-to-one).

[0139] As shown in Figure 31, 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 a plurality of arithmetic units 6200 and RAM 6201. Analysis software 6100, which analyzes the cell type of the measured cells, runs on the processor 6001. The analysis software 6100 running on the processor 6001 has the same functions as the analysis software 4832 shown in Figure 16. The analysis software 6100 analyzes the cell type of the measured cells by operating in the same manner as described in Figures 16, 17, 18, 21, 22 and related descriptions. The analysis software 6100 transmits the classification information of the measured cells to the processing unit 300 via the 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, WiFi® or Bluetooth®).

[0140] Figure 32 shows an example configuration of the processing unit 300. The processing unit 300 shown in Figure 32 does not necessarily have a parallel processing processor 3002 like the processing unit 300 shown in Figures 24 and 27. Also, the analysis software 3100 shown in Figures 24 and 27 does not necessarily have to be running on the processor 3001 shown in Figure 32. 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 a wireless communication interface (for example, WiFi or Bluetooth).

[0141] In the configurations shown in Figures 29, 30, 31, and 32, the analog signals of cells (forward scattered light signal, side scattered light signal, side fluorescence signal) generated in the FCM detection unit 410 are converted into digital signals in the A / D conversion unit 482 within the measurement unit 400. The waveform data is sent to the analysis unit 600 via the interface unit 484, bus 485, interface unit 4851, and transmission line 4852. As described above, interface unit 4851 is a dedicated interface connecting the measurement unit 400 and the analysis unit 600, providing a one-to-one connection between the two. In other words, transmission line 4852 is a transmission line that does not involve the transmission of data related to devices other than, for example, the components constituting the cell analyzer 4000 (e.g., the measurement unit 400 and the processing unit 300). Transmission line 4852 is a separate transmission line from, for example, an intranet or the internet. This makes it possible to avoid communication speed bottlenecks in digital signal transmission even when the digital signal generated by A / D conversion is performed within the measurement unit 400 and transmitted to the processing unit 300.

[0142] In the case of the cell analyzer 4000 of Configuration Example 4 shown in Figures 29, 30, 31, and 32, steps S14 (cell classification) and S15 (transmission of classification information) of the flowchart shown in Figure 20 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 the classification information is transmitted to the processing unit 300 in step S15. The processing in Figures 21 and 22 is performed by the processor 6001 and the parallel processing processor 6002 of the analysis unit 600.

[0143] (Configuration example 5) Referring to Figures 28, 33, and 34, other configuration examples of the cell analyzer 4000 will be described. This cell analyzer in Configuration Example 5 is also configured with a measurement unit 400, a processing unit 300, and an analysis unit 600, similar to Configuration Example 4 described above. The measurement unit 400 shown in Figure 33 has the same functions and configuration as the measurement unit 400 described in Figure 23 and its related descriptions. The measurement unit 400 shown in Figure 33 is connected to the analysis unit 600 via a connection cable 4202. For example, the connection cable 4202 is made of twisted pair cable and has a number of pairs of wires corresponding to the type of analog signal transmitted to the processing unit 300. The connection cable 4202 is made of a length of, for example, 1 meter or less to reduce noise during signal transmission. The measurement unit 400 transmits analog signals to the analysis unit 600 via the connection cable 4202.

[0144] The analysis unit 600 shown in Figure 34 has the same functions and configuration as the analysis unit 600 described in Figure 30 and its related descriptions. In other words, in the example of Figure 34, the parallel processing processor 6002 is mounted on the cell analyzer 4000 in a manner that is integrated into the analysis unit 600. The analysis unit 600 shown in Figure 34 further includes a connection port 6008 and an A / D converter 6009. The analog signal transmitted from the analysis unit 600 via the connection cable 4202 is input to the A / D converter 6009 via the connection port 6008. The A / D converter 6009 converts the analog signal into a digital signal by the same processing as the A / D converter 482.

[0145] The analysis unit 600 may be connected to multiple measurement units 400 via multiple connection ports 6008. If multiple measurement units 400 are provided, each measurement unit 400 may be connected to an analysis unit 600 (for example, multiple measurement units 400 and multiple analysis units 600 are connected one-to-one).

[0146] As shown in Figure 31, 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, which analyzes the cell type of the measured cells, runs on the processor 6001. The analysis software 6100 running on the processor 6001 has the same functions as the analysis software 4832 shown in Figure 16. The analysis software 6100 analyzes the cell type of the measured cells by the same operation as described in Figures 16, 17, 18, 21, 22 and related descriptions. The analysis software 6100 transmits the analysis results of the cell type of the measured cells to the processing unit 300 via the 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, WiFi or Bluetooth).

[0147] In the case of the cell analyzer 4000 of Configuration Example 5 shown in Figures 33 and 34, steps S13 (generation of digital signal), S14 (cell classification), and S15 (transmission of classification information) of the flowchart shown in Figure 20 are performed in the analysis unit 600. Specifically, the A / D conversion unit 6009 of the analysis unit 600 generates a digital signal (step S13), cell classification based on the digital signal (step S14) is performed by the processor 6001 and the parallel processing processor 6002, and the classification information is transmitted from the analysis unit 600 to the processing unit 300 (step S15). The processing in Figures 21 and 22 is performed by the processor 6001 and the parallel processing processor 6002 of the analysis unit 600.

[0148] (Data size of waveform data and digital signals) Next, the data size of the waveform data and digital signals will be described. In this embodiment, for example, the analog signal of forward scattered light (FSC), the analog signal of side scattered light (SSC), and the analog signal of side fluorescence (SFL) are sampled at multiple time points at regular intervals for a single cell. Examples of sampling rates include sampling 1024 points at 10 nanosecond intervals, sampling 128 points at 80 nanosecond intervals, or sampling 64 points at 160 nanosecond intervals. The amount of data is, for example, 2 bytes per sample. For each of FSC, SSC, and SFL, an amount of data corresponding to the sampling rate (for a rate of 1024 points, 2 bytes × 1024 = 2048 bytes) is acquired. This amount of data is the amount of data per cell. In a single measurement, for example, FSC, SSC, and SFL are measured for at least 100 cells. In addition, in a single measurement, for example, FSC, SSC, and SFL may be measured for at least 1000 cells. Furthermore, in a single measurement, FSC, SSC, and SFL may be measured for, for example, approximately 10,000 to 140,000 cells. Therefore, for example, if the number of cells measured in a single measurement is 100,000 and the sampling rate is 1024, the data size of the digital signals for each of FSC, SSC, and SFL will be 2 bytes × 1024 × 100,000 = 204,800,000 bytes, and the total for FSC, SSC, and SFL will be 614,400,000 bytes.

[0149] Furthermore, FSC, SSC, and SFL are measured for each measurement channel. Therefore, for example, if the number of cells measured in one measurement is 100,000, the sampling rate is 1024, and the number of measurement channels is 5, the data volume for each of FSC, SSC, and SFL will be 2 bytes × 1024 × 100,000 × 5 = 1,024,000,000 bytes, and the total for FSC, SSC, and SFL will be 3,072,000,000 bytes.

[0150] Based on the above, the capacity of the digital signal can range from several hundred megabytes to several gigabytes per sample, and depending on the number of cells, sampling rate, and number of measurement channels, it can be at least 1 gigabyte.

[0151] According to this embodiment, when analyzing digital signals of enormous capacity ranging from several hundred megabytes to several gigabytes per sample, the analysis processing using the deep learning algorithm 60 is completed within the cell analyzer 4000 or 4000' as described above, and the digital signals are not transmitted via the internet or intranet to an analysis server located outside the cell analyzer 4000 or 4000'. Therefore, it is possible to avoid the decrease in processing capacity that occurs due to the increased communication load when transmitting digital signals from the cell analyzer 4000 or 4000' to the analysis server.

[0152] <Second cell analyzer and measurement of biological samples using the second cell analyzer> As an example of the configuration of the second cell analyzer 4000', a block diagram is shown for a case where the measurement unit 500 is a urinary formed element analyzer or body fluid analyzer equipped with a flow cytometer for measuring a urine sample or body fluid sample.

[0153] Figure 35 is an example of a block diagram of the measurement unit 500. In Figure 35, the measurement unit 500 includes a sample distribution unit 501, a sample preparation unit 502, an optical detection unit 505, an amplification circuit 550 that amplifies the output signal of the optical detection unit 505 (the output signal amplified by the preamplifier), a filter circuit 506 that performs filtering on the output signal from the amplification circuit 550, an A / D conversion unit 507 that converts the output signal (analog signal) of the filter circuit 506 into a digital signal, a processor 4831, a parallel processing processor 4833, RAM 4834, a bus controller 4850, a storage device 511a, and a LAN (Local Area Network) adapter 512. The processor 4831, the parallel processing processor 4833, the RAM 4834, the bus controller 4850, the storage device 511a, and the LAN adapter 512 are connected to the bus 508. The deep learning algorithm 60 and the analysis software based on the deep learning algorithm 60 are stored in the storage device 511a.

[0154] The processor 4831 performs predetermined arithmetic processing on the digital signal output from the A / D converter 507, as described based on the example of the first cell analyzer 4000. The processor 4831 analyzes the waveform data using the parallel processing processor 4833. The configuration examples of the processor 4831 and the parallel processing processor 4833, and the operation of the parallel processing processor 4833 are the same as those described in Figures 11 to 19 and their related descriptions above.

[0155] The processing unit 300 is connected to the measurement unit 500 via a LAN cable, for example, through a LAN adapter 512. This processing unit 300 generates inspection results based on the measurement data acquired by the measurement unit 500. The optical detection unit 505, amplification circuit 550, filter circuit 506, A / D conversion unit 507, processor 4831, parallel processing processor 4833, and storage device 511a constitute the optical measurement unit 510, which measures the sample to be measured and generates measurement data.

[0156] Figure 36 shows the configuration of the optical detection unit 505 of the measurement unit 500. In Figure 36, the condenser lens 552 focuses the laser light emitted from the semiconductor laser light source 553, which is the light source, onto the flow cell 551, and the focusing lens 554 focuses the forward scattered light emitted from the formed elements in the measurement sample onto the forward scattered light receiving unit 555. Another focusing lens 556 focuses the side scattered light and fluorescence emitted from the formed elements onto the dichroic mirror 557. The dichroic mirror 557 reflects the side scattered light to the side scattered light receiving unit 558 and transmits the fluorescence to the fluorescence receiving unit 559. These optical signals reflect the characteristics of the formed elements in the measurement sample. The forward scattered light receiving unit 555, the side scattered light receiving unit 558, and the fluorescence receiving unit 559 then convert the optical signals into electrical signals and output the forward scattered light signal, the side scattered light signal, and the fluorescence signal, respectively. These outputs are amplified by a preamplifier and then used for processing in the next stage. Furthermore, the forward-scattered light receiving unit 555, the side-scattered light receiving unit 558, and the fluorescence receiving unit 559 can each be switched between low-sensitivity and high-sensitivity outputs by switching the drive voltage. This sensitivity switching is performed by the processor 4831, which will be described later. In this embodiment, a photodiode may be used as the forward-scattered light receiving unit 555, and a photomultiplier tube may be used as the side-scattered light receiving unit 558 and the fluorescence receiving unit 55, or a photodiode may be used as both the side-scattered light receiving unit 558 and the fluorescence receiving unit 559. The fluorescence signal output from the fluorescence receiving unit 559 is amplified by a preamplifier and then supplied to two branched signal channels. Each of the two signal channels is connected to the amplification circuit 550 described in Figure 35. Fluorescence input to one of the signal channels is amplified to high sensitivity by the amplification circuit 550.

[0157] (Preparation of the measurement sample) Figure 37 is a diagram showing the schematic functional configuration of the sample preparation unit 502 and the optical detection unit 505 shown in Figure 35. The sample distribution unit 501 shown in Figures 35 and 37 comprises a suction tube 517 and a syringe pump. The sample distribution unit 501 aspirates the sample (urine or body fluid) 00b through the suction tube 517 and dispenses it to the sample preparation unit 502. The sample preparation unit 502 comprises a reaction vessel 512u and a reaction vessel 512b. The sample distribution unit 501 distributes the quantified sample to the reaction vessel 512u and the reaction vessel 512b, respectively.

[0158] In reaction vessel 512u, the distributed biological sample is mixed with the first reagent 519u as a diluent and the third reagent 518u containing a dye. The dye in the third reagent 518u stains the formed elements in the sample. If the biological sample is urine, the sample prepared in reaction vessel 512u is used as the first measurement sample for analyzing relatively large formed elements in urine, such as red blood cells, white blood cells, epithelial cells, and tumor cells. If the biological sample is body fluid, the sample prepared in reaction vessel 512u is used as the third measurement sample for analyzing red blood cells in the body fluid.

[0159] Meanwhile, in reaction vessel 512b, the distributed biological sample is mixed with the second reagent 519b as a diluent and the fourth reagent 518b containing a dye. As will be described later, the second reagent 519b has a hemolytic effect. The dye contained in the fourth reagent 518b stains the formed elements in the sample. If the biological sample is urine, the sample prepared in reaction vessel 512b becomes the second measurement sample for analyzing bacteria in the urine. If the biological sample is body fluid, the sample prepared in reaction vessel 512b becomes the fourth measurement sample for analyzing nucleated cells (leukocytes and large cells) and bacteria in the body fluid.

[0160] A tube extends from reaction vessel 512u to the flow cell 551 of the optical detection unit 505, allowing the measurement sample prepared in reaction vessel 512u to be supplied to the flow cell 551. A solenoid valve 521u is also provided at the outlet of reaction vessel 512u. A tube also extends from reaction vessel 512b, and this tube is connected to the tube extending from reaction vessel 2u. This allows the measurement sample prepared in reaction vessel 512b to be supplied to the flow cell 551. A solenoid valve 521b is also provided at the outlet of reaction vessel 512u.

[0161] The tubes extending from reaction vessels 512u and 512b to the flow cell 551 branch off before reaching the flow cell 551, and the branched end is connected to the syringe pump 520a. A solenoid valve 521c is also provided between the syringe pump 520a and the branching point.

[0162] The tubes extending from reaction vessels 512u and 512b each branch off at the connection point, and these branches are connected to syringe pump 520b. A solenoid valve 521d is provided between the branch point and the connection point of the tube leading to syringe pump 520b.

[0163] Furthermore, a sheath liquid storage section 522, which contains the sheath liquid, is connected to the sample preparation section 502, and this sheath liquid storage section 522 is connected to the flow cell 551 by a tube. A compressor 522a is connected to the sheath liquid storage section 522, and when the compressor 522a is driven, compressed air is supplied to the sheath liquid storage section 522, and the sheath liquid is supplied from the sheath liquid storage section 522 to the flow cell 551.

[0164] Two types of suspensions (measurement samples) prepared in reaction vessels 512u and 512b are used. First, the suspension from reaction vessel 512u (the first measurement sample when the biological sample is urine; the third measurement sample when the biological sample is body fluid) is guided to the optical detection unit 505, where it forms a narrow flow surrounded by sheath fluid in the flow cell 551, and is then irradiated with laser light. Subsequently, the suspension from reaction vessel 512b (the second measurement sample when the biological sample is urine; the fourth measurement sample when the biological sample is body fluid) is guided to the optical detection unit 505, where it forms a narrow flow in the flow cell 551, and is then irradiated with laser light. This operation is performed automatically by controlling the solenoid valves 521a, 521b, 521c, 521d and the drive unit 503, etc., under the control of the processor 4831 (control unit).

[0165] The first to fourth reagents will be described in detail. The first reagent 519u is a reagent mainly composed of a buffer, and contains an osmotic pressure compensating agent to obtain a stable fluorescence signal without hemolyzing red blood cells. It is adjusted to an osmotic pressure of 100 to 600 mOsm / kg to be suitable for classification measurement. Preferably, the first reagent 519u does not have a hemolytic effect on red blood cells in urine.

[0166] Unlike the first reagent 519u, the second reagent 519b has hemolytic properties. This is to increase the permeability of the fourth reagent 518b (described later) to the bacterial cell membrane, thereby accelerating staining. Furthermore, it also helps to shrink contaminants such as mucus filaments and red blood cell fragments. The second reagent 519b contains a surfactant to achieve hemolytic properties. Various surfactants can be used, such as anionic, nonionic, and cationic surfactants, but cationic surfactants are particularly preferred. Because the surfactant can damage the bacterial cell membrane, the dye contained in the fourth reagent 518b can efficiently stain the bacterial nucleic acids. As a result, bacterial measurement can be performed with a short staining process.

[0167] In another embodiment, the second reagent 519b may not be a surfactant, but rather acquire hemolytic activity by being adjusted to an acidic or low pH. Low pH means a pH lower than that of the first reagent 19u. When the first reagent 519u is neutral or in the range of weakly acidic to weakly alkaline, the second reagent 19b is acidic or strongly acidic. When the pH of the first reagent 519u is 6.0 to 8.0, the pH of the second reagent 519b is lower than that, preferably 2.0 to 6.0.

[0168] The second reagent 519b may contain a surfactant and be adjusted to a low pH.

[0169] In yet another embodiment, the second reagent 519b may acquire a hemolytic effect by having a lower osmotic pressure than the first reagent 19u.

[0170] On the other hand, the first reagent 519u does not contain a surfactant. In other embodiments, the first reagent 519u may contain a surfactant, but the type and concentration must be adjusted so as not to cause hemolysis of red blood cells. Therefore, it is preferable that the first reagent 519u does not contain the same surfactant as the second reagent 519b, or if it does contain the same surfactant, it is preferable that the concentration is lower than that of the second reagent 519b.

[0171] Reagent 3, 518u, is a staining reagent used for the measurement of formed elements (red blood cells, white blood cells, epithelial cells, casts, etc.) in urine. The dye contained in Reagent 3, 518u, is selected to stain membranes, even those without nucleic acids. Reagent 3, 518u, preferably contains an osmotic pressure compensator to prevent red blood cell hemolysis and to obtain stable fluorescence intensity, and is adjusted to an osmotic pressure of 100-600 mOsm / kg suitable for classification measurement. Reagent 3, 18u, stains the cell membranes and nuclei (membranes) of formed elements in urine. Condensed benzene derivatives are used as staining reagents containing membrane-staining dyes; for example, cyanine dyes can be used. Reagent 3, 18u, is designed to stain not only cell membranes but also nuclear membranes. When the third reagent 518u is used, in nucleated cells such as leukocytes and epithelium, the staining intensity in the cytoplasm (cell membrane) and the staining intensity in the nucleus (nuclear membrane) combine to produce a higher staining intensity than that of nucleic acid-free formed elements in urine. This allows for the differentiation of nucleated cells such as leukocytes and epithelium from nucleic acid-free formed elements in urine such as red blood cells. As the third reagent, the reagent described in U.S. Patent No. 5891733 can be used. U.S. Patent No. 5891733 is incorporated herein by reference. The third reagent 518u is mixed with the first reagent 519u in urine or body fluid.

[0172] Reagent 4, 518b, is a staining reagent capable of accurately measuring bacteria even in specimens containing impurities of similar size to bacteria and fungi. A detailed description of Reagent 4, 518b, is provided in European Patent Publication No. 1136563. A dye that stains nucleic acids is preferably used as the dye contained in Reagent 4, 518b. As a staining reagent containing a nuclear staining dye, for example, the cyanine dye described in U.S. Patent No. 7309581 can be used. Reagent 4, 518b, is mixed with urine or the specimen together with Reagent 2, 519b. European Patent Publication No. 1136563 and U.S. Patent No. 7309581 are incorporated herein by reference.

[0173] Therefore, it is preferable that the third reagent 518u contains a dye that stains cell membranes, while the fourth reagent 518b contains a dye that stains nucleic acids. Since formed elements in urine include non-nuclear elements such as red blood cells, the inclusion of a cell membrane-staining dye in the third reagent 518u allows for the detection of formed elements in urine, including these non-nuclear elements. Furthermore, since the second reagent can damage bacterial cell membranes, the dye contained in the fourth reagent 18b allows for efficient staining of bacterial and fungal nucleic acids. As a result, bacterial measurement can be performed with a short staining procedure.

[0174] [3. First Waveform Data Analysis System] <Configuration of the first waveform data analysis system> Another embodiment relates to a waveform data analysis system. Referring to Figure 38, the first waveform data analysis system comprises a deep learning device 100A. The vendor-side device 100 operates as the deep learning device 100A. The deep learning device 100A trains a neural network 50 using training data and provides the user with a deep learning algorithm 60 trained with the training data. The deep learning algorithm 60, consisting of the trained neural network, is provided from the deep learning device 100A to the measurement unit 400 via a recording medium 98 or a communication network 99. The measurement unit 400 uses the deep learning algorithm 60, consisting of the trained neural network, to analyze the waveform data of the cells to be analyzed.

[0175] The deep learning device 100A is composed of, for example, a general-purpose computer and performs deep learning processing based on the flowchart described later. The measurement unit 400 performs waveform data analysis processing based on the flowchart described later. The recording medium 98 is a computer-readable, non-temporary, tangible recording medium such as a DVD-ROM or USB memory.

[0176] The deep learning device 100A is connected to either the measurement unit 400a or the measurement unit 500a. The configuration of the measurement unit 400a or the measurement unit 500a is the same as that of the measurement unit 400 or the measurement unit 500 described above. The deep learning device 100A acquires the training waveform data obtained by the measurement unit 400a or the measurement unit 500a. The method for generating the training waveform data is as described above.

[0177] As shown in Figure 38, the measurement unit 400a or the measurement unit 500a is equipped with flow cells 4113 and 551, respectively. The measurement unit 400a or the measurement unit 500a delivers the biological sample to the flow cells 4113 and 551. Light is irradiated onto the biological sample supplied to the flow cells 4113 and 551 from light sources 4112 and 553, and the photodetectors (4116, 4121, 4122, 555, 558, 559) detect the forward scattered light, side scattered light, and side fluorescence emitted from the cells in the biological sample. The measurement unit 400a or the measurement unit 500a generates waveform data from the forward scattered light signal, side scattered light signal, and side fluorescence signal obtained by the photodetectors (4116, 4121, 4122, 555, 558, 559) detecting the light, and transmits the waveform data to the vendor-side device 100.

[0178] <Hardware configuration of the deep learning system> Figure 39 shows an example block diagram of the vendor-side device 100 (deep learning device 100A). The vendor-side device 100 comprises a processing unit 10 (10A), an input unit 16, and an output unit 17.

[0179] The processing unit 10 includes, for example, a CPU 11 that performs data processing as described later, a memory 12 used as a work area for data processing, a storage unit 13 that records programs and processing data as described later, a bus 14 that transmits data between each unit, an interface unit (I / F unit) 15 that performs data input and output with external devices, and a GPU 19. The input unit 16 and output unit 17 are connected to the processing unit 10 via the interface unit 15. For example, the input unit 16 is an input device such as a keyboard or 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 (e.g., parallel processing) performed by the CPU 11. That is, in the following description, the processing performed by the CPU 11 also includes the processing performed by the CPU 11 using the GPU 19 as an accelerator. The GPU 19 has the same functionality as the parallel processing processor 4833 described above. Here, instead of the GPU 19, a chip suitable for neural network computation may be installed. Examples of such chips include FPGAs, ASICs, and Myriad X (Intel).

[0180] Furthermore, the processing unit 10 pre-records the deep learning algorithm 50, which consists of the program and the neural network before training according to this embodiment, in the memory unit 13, for example, in executable format, in order to perform the processing of each step described in Figure 43 below. The executable format is, for example, a format generated by converting from a programming language by a compiler. The processing unit 10 uses the program recorded in the memory unit 13 to perform the training process of the neural network 50 before training.

[0181] In the following description, unless otherwise specified, the processing performed by the processing unit 10 refers to the processing performed by the CPU 11 based on the program and neural network 50 stored in the storage unit 13 or memory 12. The CPU 11 temporarily stores necessary data (such as intermediate data during processing) in the memory 12 as a working area, and appropriately records data to be stored long-term, such as calculation results, in the storage unit 13.

[0182] <Hardware configuration of the analytical instrument> The configuration of the measurement unit 400 or 500, which processes waveform data based on an algorithm provided by the vendor-side device 100, is equivalent to the configuration described above. Alternatively, the measurement unit 400 or 500 may also incorporate the functions of the vendor-side device 100 (deep learning device 100A) and train the neural network 50 using training data. In this case, the vendor-side device 100 (deep learning device 100A) is not required.

[0183] Furthermore, the measurement unit control unit 480 pre-records the program and the deep learning algorithm 60, which consists of a trained neural network and is part of this embodiment, in the storage unit 4835, for example, in executable format, in order to perform the processing of each step described in the waveform data analysis processing below. The executable format is, for example, a format generated by converting from a programming language by a compiler. The measurement unit control unit 480 performs the processing using the program and the deep learning algorithm 60 recorded in the storage unit 4835.

[0184] As shown in Figure 40, the measurement unit control unit 480 may update the program and deep learning algorithm 60 stored in the memory unit 4835 via a communication network, for example, through a LAN adapter 481. As shown in Figure 41, the analysis unit 600 described above may transmit the digital signal and analysis results received from the measurement unit 400 to the deep learning device 100A via the interface unit 6011. The analysis unit 600 transmits waveform data and classification information corresponding to the waveform data to the deep learning device 100A, for example, via the Internet. The deep learning device 100A performs learning processing based on the waveform data transmitted from the analysis unit 600 and the classification information corresponding to the waveform data, and updates the deep learning algorithm 60. The deep learning device 100A transmits the updated deep learning algorithm 60 to the analysis unit 600. The analysis unit 600 updates the algorithm stored in the memory unit 6004 using the deep learning algorithm 60 transmitted from the deep learning device 100A. The analysis unit 600 transmits the waveform data transmitted from the measurement unit 400 to the deep learning device 100A in parallel, for example, while the waveform data transmitted from the measurement unit 400 is being processed by the processor 6001 and the parallel processing processor 6002. Once the analysis unit 600 has finished analyzing the waveform data, that is, generating the classification information, it transmits the classification information to the deep learning device 100A.

[0185] In the following description, unless otherwise specified, the processing performed by the measurement unit control unit 480 refers to the processing actually performed by the processor 4831 and the parallel processing processor 4833, based on the program and deep learning algorithm 60 stored in the memory unit 4835 or RAM 4834. For example, the processor 4831 temporarily stores necessary data (intermediate data during processing, measurement results, analysis results, etc.) in the RAM 4834 as a working area, and appropriately records data to be stored long-term, such as calculation results, in the memory unit 4835.

[0186] <Functional blocks and processing procedures> (Deep learning processing) Figure 42 shows an example of the functional blocks of a deep learning device 100A that performs deep learning. Referring to Figure 42, the processing unit 10A of the deep learning device 100A according to this embodiment includes a training data generation unit 101, a training data input unit 102, and an algorithm update unit 103. These functional blocks are realized by installing a program that causes the computer to execute deep learning processing into the storage unit 13 or memory 12 of the processing unit 10A shown in Figure 39, and then having the CPU 11 and GPU 19 execute this program. The training data database (DB) 104 and the algorithm database (DB) 105 are recorded in the storage unit 13 or memory 12 of the processing unit 10A.

[0187] The training waveform data 76a, 76b, and 76c are acquired in advance, for example, by the measurement units 400 and 500, and are stored in advance in the storage unit 13 or memory 12 of the processing unit 10A. The deep learning algorithm 50 is stored in advance in the algorithm database 105.

[0188] The processing unit 10A of the deep learning device 100A performs the processing shown in Figure 43. Using the functional blocks shown in Figure 42 as an example, steps S211, S214, and S216 shown in Figure 43 are performed by the training data generation unit 101. Step S212 is performed by the training data input unit 102. Steps S213 and S215 are performed by the algorithm update unit 103.

[0189] An example of deep learning processing performed by the processing unit 10A will be explained using Figure 43. First, the processing unit 10A acquires training waveform data 72a, 72b, and 72c. Training waveform data 72a is the waveform data of forward scattered light, training waveform data 72b is the waveform data of side scattered light, and training waveform data 72c is the waveform data of side fluorescence. The training waveform data 72a, 72b, and 72c are acquired, for example, by operator operation from the measurement units 400 and 500, from the recording medium 98, or via the I / F unit 15 via the communication network. When acquiring the training waveform data 72a, 72b, and 72c, information on which cell type the training waveform data 72a, 72b, and 72c represent is also acquired. This information on which cell type the data represents is linked to the training waveform data 72a, 72b, and 72c, and may also be input by the operator from the input unit 16.

[0190] In step S211, the processing unit 10A generates training data 75 from the acquired waveform data 72a, 72b, and 72c and the label value 77.

[0191] In step S212, the processing unit 10A inputs the training data 75 into the neural network 50 and obtains the trial results. The trial results are accumulated each time multiple training data 75 are input into the neural network 50.

[0192] In the cell type analysis method according to this embodiment, a convolutional neural network is used, and stochastic gradient descent is employed. Therefore, in step S213, the processing unit 10A determines whether a predetermined number of trial results have been accumulated. If the predetermined number of trial results have been accumulated (YES), the processing unit 10A proceeds to step S214. If the predetermined number of trial results have not been accumulated (NO), the processing unit 10A proceeds to step S215.

[0193] Next, if a predetermined number of trial results have been accumulated, in step S214, the processing unit 10A updates the connection weights w of the neural network 50 using the training results accumulated in step S212. In the cell type analysis method according to this embodiment, stochastic gradient descent is used, so the connection weights w of the neural network 50 are updated when a predetermined number of trial results have been accumulated. Specifically, the process of updating the connection weights w is a process that performs calculations using gradient descent, as shown in (Equation 12) and (Equation 13) described later.

[0194] In step S215, the processing unit 10A determines whether the neural network 50 has been trained with a specified number of training data 75. If it has been trained with a specified number of training data 75 (YES), the deep learning process is terminated.

[0195] If the neural network 50 has not been trained with the specified number of training data 75 (NO), the processing unit 10A proceeds from step S215 to step S216 and performs the processing from steps S211 to S215 for the next training waveform data.

[0196] Following the process described above, the neural network 50 is trained to obtain the deep learning algorithm 60.

[0197] (Structure of neural networks) As described above, a convolutional neural network is used in this embodiment. Figure 44(a) illustrates the structure of the neural network 50. The neural network 50 comprises an input layer 50a, an output layer 50b, and an intermediate layer 50c between the input layer 50a and the output layer 50b, with the intermediate layer 50c being composed of multiple layers. The number of layers constituting the intermediate layer 50c can be, for example, 5 or more, preferably 50 or more, and more preferably 100 or more.

[0198] In the neural network 50, multiple nodes 89 arranged in layers are connected between layers. As a result, information propagates only in one direction, as shown by arrow D in the figure, from the input layer 50a to the output layer 50b.

[0199] (Calculations at each node) Figure 44(b) is a schematic diagram showing the operations at each node. Each node 89 receives multiple inputs and calculates one output (z). In the example shown in Figure 44(b), node 89 receives four inputs. The total input (u) received by node 89 is represented, for example, by the following equation (2). Here, in this embodiment, since one-dimensional matrix data is used as the training data 75 and analysis data 85, if a variable in the operation formula corresponds to two-dimensional matrix data, a process is performed to convert the variable so that it corresponds to one-dimensional matrix data.

number

number

number

number

number

number

number

[0200] Neural network training means adjusting the weights w so that, for any input-output pair (xn, dn), the output y (xn:w) of the neural network, given an input xn, comes as close as possible to the output dn. The error function is the closeness between the function represented using the neural network and the training data.

number

number

number

[0201] If we represent each class as C1, ..., CK, then the output yK (i.e., uk) of node k in the output layer L is... (L) ) represents the probability that a given input x belongs to class CK. See (Equation 9) below. The input x is classified into the class that maximizes the probability expressed in (Equation 9).

number

[0202] Let the target output dn of the softmax function in (Equation 8) be 1 only if the output is of the correct class, and 0 otherwise. If we represent the target output in vector form dn=[dn1, ..., dnK], for example, if the correct class of input xn is C3, then only the target output dn3 will be 1, and all other target outputs will be 0. When encoded in this way, the posterior distribution is expressed in (Equation 10) below.

number

number

[0203] Minimizing the error function E(w) with respect to parameter w is equivalent to finding the local minimum of the function E(w). Parameter w is the weight of the connections between nodes. The local minimum of weight w can be found by iterative calculations that repeatedly update parameter w, starting from an arbitrary initial value. One example of such calculation is the gradient descent method.

[0204] The gradient descent method uses the vector represented by the following (Equation 12).

number

number

number

[0205] Note that the calculation using (Equation 13) may be performed on all training data (n=1, ..., N) or on only some of the training data. Gradient descent performed on only some of the training data is called stochastic gradient descent. In the cell type analysis method according to this embodiment, stochastic gradient descent is used.

[0206] [4. Building a Deep Learning Model] Blood samples collected from healthy individuals were measured as healthy blood samples, and XN CHECK Lv2 (STREC's control blood (which had undergone fixation and other processing)) was measured using a Sysmex XN-1000 as a non-healthy blood sample. FluoroCell WDF from Sysmex Corporation was used as the fluorescent staining reagent. LysaCell WDF from Sysmex Corporation was used as the hemolytic agent. For each cell in each biological sample, waveform data of forward scattered light, side scattered light, and side fluorescence were acquired at 10 nanosecond intervals from the start of forward scattered light measurement for 1024 points. For the healthy blood samples, waveform data of cells in blood collected from 8 healthy individuals were pooled as digital signals. Neutrophils (NEUT), lymphocytes (LYMPH), monocytes (MONO), eosinophils (EO), basophils (BASO), and immature granulocytes (IG) were manually classified from the waveform data of each cell, and cell type annotation (labeling) was applied to each waveform data. The measurement start point was defined as the moment when the signal intensity of forward-scattered light exceeded a threshold. Training data was generated by synchronizing the acquisition of waveform data for forward-scattered light, side-scattered light, and side-fluorescence. Control blood samples were also annotated with control blood-derived cells (CONT). The training data was input into a deep learning algorithm and trained.

[0207] Using a Sysmex XN-1000, waveform data for analysis was acquired from blood cells of healthy individuals, separate from the cell data used for training, in the same manner as the training data. This waveform data was then mixed with control blood data to create the analysis data. This analysis data was input into the constructed deep learning algorithm to obtain data for each individual cell type.

[0208] The results are shown as a mixed matrix in Figure 45. The horizontal axis shows the judgment results by the constructed deep learning algorithm, and the vertical axis shows the judgment results by humans using manual (reference method). The judgment results by the constructed deep learning algorithm showed a 98.8% agreement rate with the judgment results by the reference method, although there was some confusion between basophils and lymphocytes, and between basophils and ghosts.

[0209] Next, ROC analysis was performed for each cell type to evaluate sensitivity and specificity. Figure 46(a) shows the ROC curve for neutrophils, Figure 46(b) for lymphocytes, Figure 46(c) for monocytes, Figure 47(a) for neutrophils, Figure 47(b) for basophils, and Figure 47(c) for control blood (CONT). The sensitivity and specificity were 99.5% and 99.6% for neutrophils, 99.4% and 99.5% for lymphocytes, 98.5% and 99.9% for monocytes, 97.9% and 99.8% for eosinophils, 71.0% and 81.4% for basophils, and 99.8% and 99.6% for control blood (CONT), showing good results in all cases.

[0210] From these results, it became clear that by using a deep learning algorithm based on signals obtained from cells contained in a biological sample based on waveform data, it is possible to determine cell types with high classification accuracy.

[0211] Furthermore, when non-healthy blood cells, such as control blood, are mixed with healthy blood cells, conventional scattergram methods sometimes have difficulty determining the type of individual cells. However, by using the deep learning algorithm of this embodiment, it has been shown that it is possible to determine individual cells even when non-healthy blood cells are mixed with healthy blood cells.

[0212] [5. Analysis system using image analysis equipment] An embodiment using an image analysis device as a cell analysis device will be described. The third cell analysis device 4000'', which is an image analysis device, estimates the cell type of the imaged cells by analyzing the captured image data.

[0213] Figure 48 shows an example of the configuration of the cell analyzer 4000''. The cell analyzer 4000'' shown in Figure 48 is equipped with a measurement unit 700 and a processing unit 800, and measures and analyzes a sample 901 prepared by pretreatment by the pretreatment device 900.

[0214] The measurement unit 700 includes a flow cell 710, light sources 720 to 723, condenser lenses 730 to 733, dichroic mirrors 740 to 741, a condenser lens 750, an optical unit 751, a condenser lens 752, and an imaging unit 760. A sample 701 flows through the flow path 711 of the flow cell 710.

[0215] The light sources 720 to 723 irradiate light onto the sample 701 flowing through the flow cell 710. The light sources 720 to 723 are constituted by, for example, semiconductor laser light sources. Lights of wavelengths λ11 to λ14 are respectively emitted from the light sources 720 to 723. The condenser lenses 730 to 733 respectively condense the lights of wavelengths λ11 to λ14 emitted from the light sources 720 to 723. The dichroic mirror 740 transmits the light of wavelength λ11 and refracts the light of wavelength λ12. The dichroic mirror 741 transmits the lights of wavelengths λ11 and λ12 and refracts the light of wavelength λ13. Thus, the lights of wavelengths λ11 to λ14 are irradiated onto the sample 701 flowing through the flow path 711 of the flow cell 710. Note that the number of semiconductor laser light sources included in the measurement unit 700 is not limited as long as it is 1 or more. The number of semiconductor laser light sources can be selected, for example, from 1, 2, 3, 4, 5, or 6.

[0216] When the sample 701 flowing through the flow cell 710 is stained with a fluorescent dye, when the sample 701 is irradiated with the lights of wavelengths λ11 to λ13, fluorescence is generated from the fluorescent dye that stains the cells. For example, fluorescences of wavelengths λ21, λ22, and λ23 corresponding to wavelengths λ11, λ12, and λ13 respectively are generated. When the sample 701 flowing through the flow cell 710 is irradiated with the light of wavelength λ14, this light passes through the cells. The transmitted light of wavelength λ14 that has passed through the cells is used for generating a bright-field image.

[0217] The condenser lens 750 condenses the fluorescence generated from the sample 701 flowing through the flow path 711 of the flow cell 710 and the transmitted light that has passed through the sample 701 flowing through the flow path 711 of the flow cell 710. The optical unit 751 has, for example, a configuration in which four dichroic mirrors are combined. The four dichroic mirrors of the optical unit 751 reflect the fluorescence and the transmitted light at slightly different angles from each other and separate them on the light receiving surface of the imaging unit 760. The condenser lens 752 condenses the fluorescence and the transmitted light.

[0218] The imaging unit 760 is composed of a TDI (Time Delay Integration) camera. The imaging unit 760 can image the fluorescence and the transmitted light and output a fluorescence image corresponding to the fluorescence and a bright-field image corresponding to the transmitted light to the processing unit 800 as imaging signals.

[0219] The processing unit 800 includes, as a hardware configuration, a processing unit 811, a storage unit 812, an interface unit 816, and a bus 815. The processing unit 811, the storage unit 812, and the interface unit 816 are connected to the bus 815. Image data (for example, a fluorescence image, a bright-field image) composed of imaging signals captured by the imaging unit 760 of the measurement unit 700 is stored in the storage unit 812 via the interface unit 816. The processing unit 811 reads the image data from the storage unit 812 and analyzes the image data.

[0220] FIG. 49 shows a configuration example of the processing unit 811. The processing unit 811 includes, for example, a processor 8111, a parallel processing processor 8112, and a RAM 8113. The processor 8111, the parallel processing processor 8112, and the RAM 8113 have the same configurations and functions as the above-described processor 4831, parallel processing processor 4833, and RAM 4834, respectively. The processor 8111 and the parallel processing processor 8112 analyze the image data captured by the imaging unit 760. The parallel processing processor 8112 executes, for example, the operations of matrix data related to the image data in parallel by a plurality of arithmetic units by the processes illustrated in FIGS. 16, 17, and 18.

[0221] Furthermore, the functions of the processing unit 811 (processor 8111, parallel processing processor 8112, RAM 8113) do not necessarily have to be provided by the processing unit 800, but may be provided by the measurement unit 700.

[0222] <Generating training data> The following describes an example of generating training data in this embodiment.

[0223] Training images used to train deep learning algorithms are preferably captured in RGB and CMY color. Color images are preferably represented by 24-bit values ​​(8 bits x 3 colors) for the intensity or brightness of each primary color, such as red, green, and blue, or cyan, magenta, and yellow. A training image may contain at least one hue and its intensity or brightness, but it is more preferable to contain at least two hues and their respective intensity or brightness. Information including hue and its intensity or brightness is also called color tone.

[0224] The color information of each pixel in the training image is converted, for example, from RGB color to a format that includes luminance information and hue information. Examples of formats that include luminance information and hue information include YUV (YCbCr, YPbPr, YIQ, etc.). Here, we will explain the conversion to the YCbCr format as an example. The training image captured in RGB color is converted into image data based on luminance, image data based on a first hue (e.g., blue tones), and image data based on a second hue (e.g., red tones). The conversion from RGB to YCbCr can be performed by known methods. For example, the conversion from RGB to YCbCr can be performed according to the international standard ITU-R BT.601. The image data based on luminance, the image data based on the first hue, and the image data based on the second hue can each be represented as matrix data of grayscale values, as shown in Figure 50 (hereinafter also referred to as color matrix data 72y, 72cb, and 72cr). Image data based on luminance, image data based on the first hue, and image data based on the second hue are each represented, for example, in 256 gradations from 0 to 255. Here, instead of luminance, the first hue, and the second hue, the training images may be transformed using the three primary colors red (R), green (G), and blue (B), or the three primary colors cyan (C), magenta (M), and yellow (Y).

[0225] Next, based on the color matrix data 72y, 72cb, 72cr, a color vector data 74 is generated for each pixel, combining three tonal values: luminance 72y, first hue 72cb, and second hue 72cr.

[0226] Next, for example, if segmented neutrophils were captured in the training image, each color vector data 74 generated from the training image will be assigned a label value 77 of "1" to indicate that it is a segmented neutrophil, and this will become the training data 75. In Figure 50, for convenience, the training data 75 is represented as 3 pixels × 3 pixels, but in reality, there is color vector data for each pixel in the training image that was captured.

[0227] Figure 51 shows an example of label value 77. Different label values ​​77 are assigned depending on the cell type and the presence or absence of characteristics of each cell.

[0228] <Overview of Deep Learning> Figure 50 is used as an example to explain the overview of neural network training. The neural network 50 is preferably a convolutional neural network. The number of nodes in the input layer 50a of the neural network 50 corresponds to the product of the number of pixels in the input training data 75 and the number of luminance and hue values ​​contained in the image (for example, in the above example, luminance 72y, first hue 72cb, and second hue 72cr). The color vector data 74 is input to the input layer 50a of the neural network 50 as a set 76. The label value 77 of each pixel in the training data 75 is used as the output layer 50b of the neural network to train it.

[0229] The neural network 50 extracts features regarding morphological cell types and cell characteristics based on the training data 75. The output layer 50b of the neural network outputs results that reflect these features.

[0230] In Figure 50, the label 50c indicates the intermediate layer.

[0231] The deep learning algorithm 60 having the thus trained neural network 60 is used as a classifier to identify which of several cell types, which belong to a predetermined cell group and are morphologically classified, the cell being analyzed belongs to.

[0232] <Image Analysis Method> Figure 52 shows an example of an image analysis method. In this image analysis method, analysis data 81 is generated from an analysis image of the cells to be analyzed. The analysis image is an image of the cells to be analyzed.

[0233] For example, in this embodiment, imaging in the imaging device is preferably performed using RGB color and CMY color, etc. The color image is preferably represented by 24-bit values ​​(8 bits x 3 colors) for the intensity or brightness of each primary color, such as red, green, and blue, or cyan, magenta, and yellow. The analysis image may include at least one hue and its intensity or brightness, but it is more preferable to include at least two hues and their respective intensity or brightness. Information including a hue and its intensity or brightness is also called color tone.

[0234] For example, RGB color can be converted to a format that includes luminance and hue information. Examples of formats that include luminance and hue information include YUV (YCbCr, YPbPr, YIQ, etc.). Here, we will explain the conversion to the YCbCr format as an example. Here, the RGB color training image is converted to image data based on luminance, image data based on a first hue (e.g., blue tones), and image data based on a second hue (e.g., red tones). The conversion from RGB to YCbCr can be performed by known methods. For example, the conversion from RGB to YCbCr can be performed according to the international standard ITU-R BT.601. The image data corresponding to luminance, the first hue, and the second hue can each be represented as matrix data of gradation values, as shown in Figure 52 (hereinafter also referred to as color matrix data 79y, 79cb, and 79cr). Luminance, the first hue, and the second hue 72Cr are each represented in 256 gradations from 0 to 255. Here, instead of luminance, first hue, and second hue, the training images may be transformed using the three primary colors of red (R), green (G), and blue (B), or the three primary colors of cyan (C), magenta (M), and yellow (Y).

[0235] Next, based on the color matrix 79y, 79cb, 79cr, color vector data 80 is generated for each pixel, combining three tonal values: luminance 79y, first hue 79cb, and second hue 79cr. The set of color vector data 80 generated from one analysis image is then generated as analysis data 81.

[0236] It is preferable that the generation of the analysis data 81 and the generation of the training data 75 make at least the imaging conditions and the generation conditions of the vector data input to the neural network from each image the same. 2]]

[0237] The analysis data 81 is input to the input layer 60a of the neural network 60 that constitutes the trained deep learning algorithm 60. The deep learning algorithm extracts features from the analysis data 81 and outputs the result from the output layer 60b of the neural network 60. The value output from the output layer 60b is the probability that the cell to be analyzed included in the analysis image belongs to each of the morphological cell classifications and features input as the training data.

[0238] <00009(END]]Among these probabilities, it is determined that the cell to be analyzed included in the analysis image belongs to the morphological classification with the highest value, and a label value associated with the morphological cell type or cell feature is output. The label value itself, or data obtained by replacing the label value with information indicating the presence or absence of the morphological cell type or cell feature (e.g., terms, etc.) is output as the analysis result 83 regarding the cell morphology. In FIG. 52, from the analysis data 81, the label value "1" is output as the most likely label value 82 by the discriminator, and the character data "segmented neutrophil" corresponding to this label value is output as the analysis result 83 regarding the cell morphology.

[0239] Reference numeral 60c in FIG. 52 indicates the intermediate layer.

[0240] [6. Other Forms] As described above, the present invention has been described by way of an overview and specific embodiments, but the present invention is not limited to the above-described overview and each embodiment.

[0241] ​​In the above embodiment, the training data generation unit 101, training data input unit 102, algorithm update unit 103, analysis data generation unit 201, analysis data input unit 202, and analysis unit 203 are executed on a single processor 4831 and a single parallel processing processor 4833. However, these functional blocks do not necessarily need to be executed on a single processor and a parallel processing processor, and may be executed in a distributed manner across multiple processors and multiple parallel processing processors.

[0242] In the above embodiment, a program for performing each step described in Figure 43 is pre-recorded in the storage unit 4835. Alternatively, the program may be installed in the measurement unit control unit 480 from a computer-readable, non-temporary tangible recording medium 98, such as a DVD-ROM or USB memory. Or, the measurement unit control unit 480 may be connected to a communication network 99, and the program may be downloaded and installed from, for example, an external server (not shown) via the communication network 99.

[0243] Figure 53 shows one embodiment of the analysis results. Figure 53 shows the cell types and the number of cells of each type contained in the biological sample measured by flow cytometry, with the label values ​​shown in Figure 4 assigned. Instead of displaying the number of cells, or along with the display of the number of cells, the proportion (e.g., %) of each cell type to the total number of cells counted may be output. The cell count can be determined by using the number of label values ​​corresponding to each output cell type (the number of identical label values) as a coefficient. In addition, the output results may include a warning indicating that abnormal cells are present in the biological sample. Figure 53 shows an example where an exclamation mark is added as a warning in the abnormal cell section, but this is not limited to this. Furthermore, the distribution of each cell tumor may be output as a scattergram. When outputting as a scattergram, for example, the highest value obtained when acquiring signal intensity can be plotted, for example, with lateral fluorescence intensity on the vertical axis and lateral scattered light intensity on the horizontal axis.

[0244] [Additional Notes] The present invention also includes the following embodiments.

[0245] [Aspect 1] A sample preparation unit that mixes the sample and reagent to prepare a measurement sample, A detection unit that passes the measurement sample through a flow cell and acquires matrix data whose elements are values ​​indicating the signal intensity of each of the multiple cells contained in the measurement sample that pass through the flow cell at multiple time points, A parallel processing processor that performs parallel processing of matrix operations included in a deep learning algorithm that outputs information about the cell type for each of the multiple cells in response to the input matrix data, A processor that causes the sample preparation unit to prepare the measurement sample, the detection unit to acquire the matrix data, and the parallel processing processor to execute the parallel processing, A cell analysis device equipped with the following features.

[0246] [Aspect 2] The detection unit comprises a light source that irradiates light onto the cells passing through the flow cell, and a light receiving unit that receives light emitted from the cells, and acquires matrix data whose elements are values ​​indicating the signal intensity at multiple time points obtained by the light receiving unit receiving light emitted from the cells passing through the flow cell. The cell analyzer described in Embodiment 1.

[0247] [Aspect 2] The light-receiving unit comprises a first light-receiving unit that receives a first type of light from the cell and a second light-receiving unit that receives a second type of light from the cell. The aforementioned matrix data is (1) First data comprising values ​​indicating the signal intensity at multiple time points obtained by the first light receiving unit receiving the first type of light emitted from cells passing through the flow cell, and (2) The second data includes values ​​representing the signal intensity at multiple time points obtained by the second light receiving unit receiving the second type of light emitted from cells passing through the flow cell, The cell analyzer described in Embodiment 2.

[0248] [Aspect 4] The deep learning algorithm determines the cell type of the cell corresponding to the input matrix data from among a plurality of cell types, including monocytes, neutrophils, lymphocytes, eosinophils, basophils, and abnormal cells. The cell analyzer described in Embodiment 1.

[0249] [Aspect 5] The matrix data is transmitted to the parallel processing processor via a transmission path provided by the cell analysis device. The cell analyzer described in Embodiment 1.

[0250] [Aspect 6] The cell analysis apparatus according to embodiment 1, wherein the matrix data is transmitted to the parallel processing processor via a transmission path different from the internet or intranet.

[0251] [Aspect 7] The cell analysis apparatus according to embodiment 5, wherein the transmission line has a communication bandwidth of 1 gigabit / second or more.

[0252] [Aspect 8] The cell analysis apparatus according to embodiment 5, wherein the transmission path is a bus, and the matrix data is transmitted to the parallel processing processor via the bus.

[0253] [Aspect 9] The matrix data converted from the analog signals relating to the plurality of cells measured by the flow cytometer is transmitted to the parallel processing processor via the transmission path. A cell analyzer according to any one of embodiments 5 to 7.

[0254] [Aspect 10] The information includes an identifier for identifying the cell type, A cell analyzer according to any one of embodiments 1 to 9.

[0255] [Aspect 11] The information includes the probability that the cell belongs to each of the plurality of cell types. A cell analyzer according to any one of embodiments 1 to 10.

[0256] [Aspect 12] The analysis results, including an identifier for identifying the cell type, are transmitted to a processing unit that performs the analysis of the analysis results. A cell analyzer according to any one of embodiments 1 to 11.

[0257] [Aspect 13] The analysis results, including the probability that the cells belong to each of the multiple cell types, are transmitted to a processing unit that performs analysis of the analysis results. A cell analyzer according to any one of embodiments 1 to 12.

[0258] [Aspect 14] The parallel processing processor executes, as part of the parallel processing, a plurality of arithmetic operations related to the analysis of the matrix data in parallel. A cell analyzer according to any one of embodiments 1 to 13.

[0259] [Aspect 15] The aforementioned parallel processing processor is The system has multiple computing units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of embodiments 1 to 14.

[0260] [Aspect 16] The parallel processing processor performs, as part of the parallel processing, a filtering process for extracting features from the matrix data in parallel. A cell analyzer according to any one of embodiments 1 to 15.

[0261] [Aspect 17] The parallel processing processor performs, as parallel processing, multiple operations in the convolutional layer of the deep learning algorithm in parallel. A cell analyzer according to any one of embodiments 1 to 16.

[0262] [Aspect 18] The parallel processing processor executes the parallel processing according to a single instruction. A cell analyzer according to any one of embodiments 1 to 17.

[0263] [Aspect 19] The matrix data is data based on signals detected by irradiating the cells with light. A cell analyzer according to any one of embodiments 1 to 18.

[0264] [Aspect 20] The sampled data obtained by sampling the signals from the cells at a predetermined rate is input to the parallel processing processor as matrix data. The parallel processing processor analyzes the sampled data by performing the parallel processing. A cell analyzer according to any one of embodiments 1 to 19.

[0265] [Aspect 21] The image data obtained by irradiating the cells with light is input to the parallel processing processor as matrix data. The parallel processing processor analyzes the image data by performing the parallel processing. A cell analyzer according to any one of embodiments 1 to 20.

[0266] [Aspect 22] The parallel processing processor performs at least 100 matrix operations for each of the cells. A cell analyzer according to any one of embodiments 1 to 21.

[0267] [Aspect 23] The parallel processing processor performs at least 1000 matrix operations for each of the cells. A cell analyzer according to any one of embodiments 1 to 22.

[0268] [Aspect 24] The parallel processing processor analyzes the matrix data corresponding to each of the at least 100 cells. A cell analyzer according to any one of embodiments 1 to 23.

[0269] [Pattern 25] The parallel processing processor analyzes the matrix data corresponding to each of the at least 1000 cells. A cell analyzer according to any one of embodiments 1 to 24.

[0270] [Aspect 26] Each of the parallel processing processors analyzes the matrix data having a capacity of at least 1 gigabyte. A cell analyzer according to any one of embodiments 1 to 25.

[0271] [Aspect 27] The aforementioned parallel processing processor is The system has at least 10 arithmetic units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of embodiments 1 to 26.

[0272] [Aspect 28] The aforementioned parallel processing processor is The system has at least 100 arithmetic units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of embodiments 1 to 27.

[0273] [Aspect 29] The aforementioned parallel processing processor is The system has at least 1000 computing units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of embodiments 1 to 28.

[0274] [Aspect 30] The parallel processing processor takes the matrix data read from memory having a capacity of at least 1 gigabyte as input and performs the parallel processing. A cell analyzer according to any one of embodiments 1 to 29.

[0275] [Aspect 31] A cell analyzer according to any one of embodiments 1 to 30, wherein the matrix data relating to cells in a sample is a combination of matrix data based on multiple types of signals obtained from cells.

[0276] [Aspect 32] In cell analysis equipment, Mix the sample and reagent to prepare the measurement sample. The measurement sample is passed through a flow cell, and matrix data is obtained whose elements are values ​​representing the signal intensity of each of the multiple cells contained in the measurement sample that pass through the flow cell at multiple time points. A parallel processing processor performs parallel processing of matrix operations included in a deep learning algorithm that outputs cell type information for each of the multiple cells in response to the input matrix data. The processor is used to perform the preparation of the measurement sample, acquire the matrix data, and cause the parallel processing processor to perform the parallel processing. A cell analysis method that includes the following. [Explanation of Symbols]

[0277] 50 Deep learning algorithms before training 60 pre-trained deep learning algorithms 400, 400a, 500, 500a, 700 measurement units 410 FCM detection unit 450 Specimen aspiration section 482, 507, 3008, 6009 A / D conversion unit 3001, 4831, 6001, 8111 Processors (Host Processors) 3002, 4833, 6002, 8112 parallel processing processors 4000, 4000', 4000'' cell analyzer 3200, 6200, 4836 arithmetic units

Claims

1. A measurement unit comprising a flow cytometer and configured to measure multiple cells contained in a sample, A first processor configured to perform information processing related to the analysis of the plurality of cells, A second processor configured to perform parallel processing, Equipped with, The measurement unit flows the sample through the flow cell of the flow cytometer and acquires an analog signal corresponding to the intensity of light emitted from the irradiated cells for each of the multiple cells, thereby acquiring matrix data whose elements are values ​​that digitally represent the analog signal levels of each of the multiple cells at multiple time points. The second processor is configured to perform parallel processing for processing matrix data according to an artificial intelligence algorithm that includes multiple matrix operations. The first processor is configured to classify the cell type of each of the multiple cells based on the results of the parallel processing performed by the second processor. Cell analysis device.

2. The flow cytometer includes a light source for irradiating the flow cell with light, and a detection unit for detecting the light emitted from each of the cells by the light irradiated by the light source. The cell analyzer according to claim 1.

3. The detection unit is configured to detect multiple types of light emitted from each of the cells. The cell analyzer according to claim 2.

4. The aforementioned artificial intelligence algorithm is a deep learning algorithm. A cell analyzer according to any one of claims 1 to 3.

5. The deep learning algorithm determines the cell type of the cell corresponding to the input matrix data from among a plurality of cell types, including monocytes, neutrophils, lymphocytes, eosinophils, basophils, and abnormal cells. The cell analyzer according to claim 4.

6. The matrix data is transmitted to the second processor via the transmission path provided by the cell analysis device. A cell analyzer according to any one of claims 1 to 5.

7. The matrix data is transmitted to the second processor via a transmission path different from the internet or intranet. The cell analyzer according to claim 6.

8. The aforementioned transmission line has a communication bandwidth of 1 gigabit / second or more. The cell analyzer according to claim 6 or 7.

9. The transmission path is a bus, and the matrix data is transmitted to the second processor via the bus. A cell analyzer according to any one of claims 6 to 8.

10. The measurement unit includes an A / D converter configured to convert the analog signals for each of the plurality of cells measured by the flow cytometer into digital signals. The A / D conversion unit obtains the matrix data of the cell as waveform data corresponding to the cell by sampling each analog signal at a predetermined sampling rate. A cell analyzer according to any one of claims 1 to 9.

11. To generate information including an identifier for identifying the aforementioned cell type, A cell analyzer according to any one of claims 1 to 10.

12. The system generates information including the probability that the cell belongs to each of the multiple cell types. A cell analyzer according to any one of claims 1 to 11.

13. The analysis results, including an identifier for identifying the cell type, are transmitted to a processing unit that performs the analysis of the analysis results. A cell analyzer according to any one of claims 1 to 12.

14. The analysis results, including the probability that the cells belong to each of the multiple cell types, are transmitted to a processing unit that performs analysis of the analysis results. A cell analyzer according to any one of claims 1 to 13.

15. The second processor, as part of the parallel processing, executes in parallel a plurality of arithmetic operations related to the analysis of the matrix data. A cell analyzer according to any one of claims 1 to 14.

16. The second processor is, The system has multiple computing units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of claims 1 to 15.

17. The second processor, as part of the parallel processing, executes a filtering process in parallel to extract features from the matrix data. A cell analyzer according to any one of claims 1 to 16.

18. The artificial intelligence algorithm is a deep learning algorithm, and the second processor executes, as parallel processing, multiple operations in the convolutional layer of the deep learning algorithm in parallel. A cell analyzer according to any one of claims 1 to 17.

19. The second processor executes the parallel processing according to a single instruction. A cell analyzer according to any one of claims 1 to 18.

20. The matrix data is data based on signals detected by irradiating the cells with light. A cell analyzer according to any one of claims 1 to 19.

21. The signals obtained by measuring the aforementioned cells are sampled at a predetermined rate, and the resulting sampled data is input to the second processor as matrix data. The second processor analyzes the sampled data by performing the parallel processing. A cell analyzer according to any one of claims 1 to 20.

22. The image data obtained by irradiating the cells with light is input to the second processor as matrix data. The second processor analyzes the image data by performing the parallel processing. A cell analyzer according to any one of claims 1 to 21.

23. The second processor performs at least 100 matrix operations for each of the cells. A cell analyzer according to any one of claims 1 to 22.

24. The second processor performs at least 1000 matrix operations for each of the cells. A cell analyzer according to any one of claims 1 to 23.

25. The second processor analyzes the matrix data corresponding to each of at least 100 of the cells. A cell analyzer according to any one of claims 1 to 24.

26. The second processor analyzes the matrix data corresponding to each of the at least 1,000 cells. A cell analyzer according to any one of claims 1 to 25.

27. Each of the second processors analyzes the matrix data having a capacity of at least 1 gigabyte. A cell analyzer according to any one of claims 1 to 26.

28. The second processor is, The system has at least 10 arithmetic units capable of performing calculations related to the analysis of the matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of claims 1 to 27.

29. The second processor is, The system has at least 100 arithmetic units capable of performing calculations related to the analysis of the matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of claims 1 to 28.

30. The second processor is, The system has at least 1000 arithmetic units capable of performing calculations related to the analysis of the aforementioned matrix data, As the parallel processing, the calculation processing performed by each of the calculation units is executed in parallel. A cell analyzer according to any one of claims 1 to 29.

31. The second processor takes the matrix data read from memory having a capacity of at least 1 gigabyte as input and executes the parallel processing. A cell analyzer according to any one of claims 1 to 30.

32. The matrix data concerning cells in a sample is a combination of matrix data based on multiple types of signals obtained from the cells. A cell analyzer according to any one of claims 1 to 31.

33. A cell analysis method using a cell analysis device including a first processor, a second processor, and a flow cytometer, By flowing the sample through the flow cell of the flow cytometer and acquiring an analog signal corresponding to the intensity of light emitted from the irradiated cells for each of the multiple cells, matrix data for each of the multiple cells in the sample is obtained. Here, the matrix data has as elements values ​​that digitally represent the analog signal levels at multiple points in time. The second processor performs parallel processing to process the matrix data according to an artificial intelligence algorithm that includes multiple matrix operations. The first processor classifies the cell type of each of the multiple cells based on the results of parallel processing by the second processor. A cell analysis method that includes this.

Images