WO2015157970A1 - 一种流式细胞分析仪及其多维数据分类方法、装置 - Google Patents

一种流式细胞分析仪及其多维数据分类方法、装置 Download PDF

Info

Publication number
WO2015157970A1
WO2015157970A1 PCT/CN2014/075613 CN2014075613W WO2015157970A1 WO 2015157970 A1 WO2015157970 A1 WO 2015157970A1 CN 2014075613 W CN2014075613 W CN 2014075613W WO 2015157970 A1 WO2015157970 A1 WO 2015157970A1
Authority
WO
WIPO (PCT)
Prior art keywords
parameter
cell population
auxiliary
interest
particle
Prior art date
Application number
PCT/CN2014/075613
Other languages
English (en)
French (fr)
Inventor
徐燕
钱程
李鑫
谢俊斌
Original Assignee
深圳迈瑞生物医疗电子股份有限公司
北京深迈瑞医疗电子技术研究院有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳迈瑞生物医疗电子股份有限公司, 北京深迈瑞医疗电子技术研究院有限公司 filed Critical 深圳迈瑞生物医疗电子股份有限公司
Priority to CN201480074432.5A priority Critical patent/CN105940301B/zh
Priority to PCT/CN2014/075613 priority patent/WO2015157970A1/zh
Publication of WO2015157970A1 publication Critical patent/WO2015157970A1/zh
Priority to US15/295,891 priority patent/US20170102310A1/en

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N15/00Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
    • G01N15/06Investigating concentration of particle suspensions
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/483Physical analysis of biological material
    • G01N33/487Physical analysis of biological material of liquid biological material
    • G01N33/49Blood
    • G01N33/4915Blood using flow cells
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N15/00Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
    • G01N15/10Investigating individual particles
    • G01N15/14Electro-optical investigation, e.g. flow cytometers
    • G01N15/1429Electro-optical investigation, e.g. flow cytometers using an analyser being characterised by its signal processing
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N15/00Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
    • G01N15/10Investigating individual particles
    • G01N15/14Electro-optical investigation, e.g. flow cytometers
    • G01N15/1456Electro-optical investigation, e.g. flow cytometers without spatial resolution of the texture or inner structure of the particle, e.g. processing of pulse signals
    • G01N15/1459Electro-optical investigation, e.g. flow cytometers without spatial resolution of the texture or inner structure of the particle, e.g. processing of pulse signals the analysis being performed on a sample stream
    • G01N15/075
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N15/00Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
    • G01N15/10Investigating individual particles
    • G01N2015/1006Investigating individual particles for cytology
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N15/00Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
    • G01N15/10Investigating individual particles
    • G01N15/14Electro-optical investigation, e.g. flow cytometers
    • G01N2015/1402Data analysis by thresholding or gating operations performed on the acquired signals or stored data

Definitions

  • the present application relates to the field of cell analysis, and in particular to a flow cell analyzer and a multi-dimensional data automatic classification method and device thereof. Background technique
  • the flow cytometer analyzes and recognizes the cells by receiving various light signals from the cells in the flow after laser irradiation.
  • the flow optical signals usually include forward scattered light (FSC), side scattered light (SSC), and various kinds of fluorescence. (FL1, FL2) These signals form different parameters of the streaming data, or channels, or dimensions. These light signals reflect the physicochemical characteristics of the cells or microspheres, such as size, particle size, and labeled fluorescein.
  • the flow cytometer collects the light signals of each channel, and analyzes the cells using the gated technique.
  • the gates need to specify a range of target cell populations in certain dimensions for analysis.
  • Manually setting the door is based on the subjective judgment of the person. The results of different people have certain differences, and it is difficult to achieve consistency of results.
  • Computer technology facilitates data analysis for flow cytometry.
  • commercial manufacturers provide automatic door-grid functions. The important advantage of this function is that it can not only reduce the workload of people, but also Reduce the error caused by the subjective judgment process of manually setting the door and improve the consistency of the analysis results.
  • Another advantage of automatic door setting is that multiple parameters can be analyzed simultaneously to obtain more information and improve the accuracy of the door.
  • the present application provides a flow multi-dimensional data automatic classification method, comprising: acquiring particle characteristic data for characterizing a cell particle, wherein the particle feature data is a plurality of channels through a flow cytometer Collecting the data set; determining at least one auxiliary parameter according to the detection item, wherein each auxiliary parameter is one-dimensional in the particle characteristic data; performing statistics on the particle characteristic data based on the auxiliary parameter; extracting the cell of interest from the statistical result of the auxiliary parameter Group; the particle feature data is counted based on the main parameter, and the main parameter is the final pass
  • the method of setting the door is to circle the parameter of the target cell group in the statistical result based on the parameter, which is a dimension different from the auxiliary parameter in the particle feature data; and map the extracted cell population of interest to the statistical result of the main parameter;
  • the target cell population is obtained by using the distribution position and edge of the cell population of interest and combining the main parameters to set the gate.
  • the present application provides a streaming multi-dimensional data automatic classification device, including: a data acquisition unit, configured to acquire particle characteristic data for characterizing a cell particle, wherein the particle characteristic data is through a flow cell a data set collected by the plurality of channels of the analyzer; an auxiliary parameter determining unit, configured to determine at least one auxiliary parameter according to the detected item, wherein each auxiliary parameter is one-dimensional in the particle characteristic data; and the auxiliary parameter statistical unit is configured to The particle feature data is calculated based on the auxiliary parameter; the first extracting unit is configured to extract the cell population of interest from the statistical result of the auxiliary parameter; the main parameter statistical unit is configured to perform statistics on the particle feature data based on the main parameter, the main parameter In order to finally circle the parameter of the target cell group based on the statistical result of the parameter, which is a dimension different from the auxiliary parameter in the particle feature data; a mapping unit for mapping the extracted cell population of interest to The statistical result of the main parameter; the second extraction unit, for a data acquisition unit, configured to acquire particle characteristic
  • the present application provides a flow cytometer, comprising: an optical detecting device, configured to perform light irradiation on a sample to be measured, collect light information generated by light irradiation of a particle, and output the particle The particle characteristic data corresponding to the optical information; the data processing device, configured to receive the particle characteristic data, and process the particle characteristic data, wherein the processing device comprises the above-mentioned streaming multi-dimensional data automatic sorting device.
  • Figure 1 is a schematic diagram of the principle of a flow cytometer
  • Figure 3 is a flow chart for extracting a population of cells of interest
  • Figure 4 is a flow chart for finding a target cell population using the distribution position and edge of the cell population of interest
  • FIG. 5 is a schematic structural diagram of a stream type multi-dimensional data automatic sorting apparatus
  • 6a-6k are diagrams of respective processing results of a multidimensional data classification method
  • Figure 7 is a diagram showing the result of processing a cell population of interest
  • Figure 8 is a diagram showing the result of processing a cell population of interest
  • Figure 9a is a diagram showing the results of processing a cell population of interest
  • Fig. 9b is a processing result of gate setting based on the automatic multi-dimensional data classification method. detailed description
  • the embodiment of the present application provides a flow cytometer, please refer to FIG. Schematic diagram of a cell analyzer, the flow cell analyzer includes an optical detection device 20, a delivery device 30, and a data processing device 40.
  • the delivery device 30 is used to deliver sample fluid into the optical detection device 20.
  • the delivery device 30 typically includes a delivery line and a control port that is delivered to the optical inspection device 20 through a delivery line and a control port.
  • the optical detecting device 20 is configured to illuminate a sample liquid flowing through a detection area thereof, and collect cells through a plurality of channels (cells are very small particles, and thus cells are also referred to as particles).
  • Various kinds of light information generated by light irradiation (for example, scattered light information and/or fluorescent information), and converted into corresponding electrical signals, which correspond to the characteristics of the particles, and become particle characteristic data, that is, each cell particle is represented by a set of parameters of multiple dimensions, the data
  • the set can be represented as an array, for example particle A is characterized by an array A ( Al, A2, ..., Ai ).
  • the optical detecting device 20 may include a light source 1025, a flow chamber 1022 as a detection area, a light collecting device 1023 disposed on the optical axis and/or a side of the optical axis, and a photosensor 1024.
  • the sample liquid is irradiated to the detection area 1021 by the flow chamber 1022 of the detection area under the sheath of the sheath liquid, and each of the cell particles in the sample liquid emits scattered light (or scattered light and fluorescence) after being irradiated by the light beam.
  • the light collecting device 1023 collects and shapes the scattered light (or the scattered light and the fluorescent light), and collects the shaped light to be irradiated to the photoelectric sensor 1024, and the photoelectric sensor
  • the data processing device 40 is configured to perform analysis processing on the feature data of the received particles. Please refer to Figure 2, which is a flow chart for processing the feature data of the particle, including the following steps:
  • Step 101 Obtain particle feature data.
  • particle characteristic data is used to characterize cell particles.
  • Particle characterization data is a collection of data collected through multiple channels of a flow cytometer.
  • Step 102 Determine an auxiliary parameter.
  • the auxiliary parameter is defined relative to the main parameter.
  • the main parameter refers to the parameter that finally sets the target cell group by the method of setting the door.
  • the main parameter is usually determined according to the detection item, that is, the particle characteristic data is counted based on the selected parameter, for example, generating a histogram. Graph or scatter plot, in the statistical results, the target cell population is determined by gated method, and the parameter becomes the main parameter of the target cell population.
  • Auxiliary parameters are parameters that can help the primary parameter locate the target cell population or distinguish the interfering cell population.
  • the auxiliary parameter can be selected according to the antibody and experience used in the detection item. For example, a comparison table between the detection item and the auxiliary parameter can be determined in advance. In a preferred embodiment, when the auxiliary parameter is determined, the detection item can be checked according to the detection item. The table is obtained.
  • the target cells or interfering cells are specifically expressed.
  • the parameter serves as an auxiliary parameter, for example, the parameter value of the target cell or the interfering cell in the parameter item is significantly different from the parameter value of the other cell in the parameter item or has obvious characteristics.
  • Each primary parameter is one dimension in the particle characterization data
  • each auxiliary parameter is one dimension of the particle characterization data that is different from the primary parameter.
  • the auxiliary parameter may be one or more, and the auxiliary parameter may be determined according to the detection item.
  • Step 103 Perform statistics on the particle feature data based on the auxiliary parameters.
  • the statistics of the particle feature data based on the auxiliary parameters may be based on a single auxiliary parameter.
  • the auxiliary parameter is Al, A2, ..., the nth dimension data in Ai, and the nth dimension of all the cell particles.
  • the data A ( An ) is statistically analyzed to form a one-dimensional statistical chart, such as a histogram.
  • statistics on the particle feature data based on the auxiliary parameters may be based on a combination of the auxiliary parameters and other parameters, or may be based on a combination of a plurality of auxiliary parameters.
  • the first and n-dimensional data A (Al , An ) of all the cell particles are counted to form a two-dimensional statistical chart, such as a scatter plot.
  • Step 104 Extract a cell population of interest from the statistical result of the auxiliary parameter.
  • the cell population of interest can be extracted from the statistical result of the auxiliary parameter according to the specificity of the detection item and the target cell population or the interfering cell population on the auxiliary parameter.
  • the cell population of interest is used to assist in locating the location and edge of the target cell population, which may be the final target cell population or a portion thereof, or may be interfering cells. Since the specificity of the cell population of interest in the auxiliary parameters has been taken into consideration when selecting the auxiliary parameters, the cell population can be first divided in the statistical results of the auxiliary parameters, and then based on the detection items and the cell population of interest on the auxiliary parameters.
  • the distribution characteristics determine the cell population that conforms to the specificity as the cell population of interest, for example, determining the cell population with the maximum, minimum, or within the set parameter value as the cell population of interest.
  • the process of extracting the cell population of interest is as shown in FIG. 3, and includes the following steps: Step 1041: performing a threshold processing on the particle feature data based on the auxiliary parameter, and the threshold value is to remove the image.
  • Step 1041 performing a threshold processing on the particle feature data based on the auxiliary parameter, and the threshold value is to remove the image.
  • the point where the gray value is smaller than the threshold value to remove the interference, and the image can be changed into a binary image by the threshold processing to facilitate subsequent processing.
  • Step 1042 Perform a communication area labeling on the threshold processed image, and mark the cell labeled as a connected area as a cell community.
  • Step 1043 Find a center of each communication area, and use an auxiliary parameter value at the center of the communication area as an auxiliary parameter value of the cell community.
  • step 1044 the population of cells of interest is determined. According to the distribution characteristics of the cell population of interest on the auxiliary parameters and the auxiliary parameter values of each cell population, the cell population conforming to the specific expression was determined as the cell population of interest.
  • Step 105 Perform statistics on the particle feature data based on the main parameters. Similarly, when all cell particles are counted based on the main parameters, statistics can be performed based on a single main parameter to form a histogram, or a combination of main parameters and other parameters can be used to form a two-dimensional or multi-dimensional scatter plot. .
  • Step 106 Map the cell population of interest into the statistical result of the main parameter.
  • the cell particles belonging to the cell population of interest are marked.
  • map each cell population of interest to the statistical results of the main parameters.
  • Step 107 Find a target cell population.
  • the target cell population is obtained by using the distribution position and edge of the cell population of interest and combining the main parameters to set the gate.
  • the watershed algorithm, clustering algorithm, contour method and/or gradient method can be used to find the boundary between the target cell group and other cells, and the target cell population can be obtained by gated method.
  • the statistical result of the main parameter is a scattergram
  • the cell population of interest belongs to a part of the target cell group
  • the target cell population is obtained by using the distribution position and the edge of the cell population of interest.
  • the distribution area of the cell population of interest is taken as a foreground.
  • the method of segmenting the foreground and the background includes a water cooling algorithm, an active contour algorithm or a random walk algorithm.
  • the method further includes: performing polygon approximation processing on the boundary to obtain a polygon gate, and using the cells in the gate as the target cell group.
  • the above step 105 may also be performed before the auxiliary parameter or synchronized with the auxiliary parameter.
  • the auxiliary parameter and the main parameter of the target cell group of each detection item are selected, and the particle characteristic data of the cell is respectively counted based on the auxiliary parameter and the main parameter, and the cell of interest is obtained from the statistical result of the auxiliary parameter.
  • the group maps the cell population of interest to the statistical results based on the main parameters, and finally uses the distribution position and edge of the cell population of interest, and combines the main parameters to set the gate to obtain the target cell population.
  • the cell population of interest belongs to a part of the target cell population, the distribution of the cell population of interest in the statistical results of the main parameters, and the location and edge determination of the target cell population
  • the distribution position and edge of the target cell population can be determined according to the distribution position and edge of the cell population of interest.
  • the cell population of interest is a cell that interferes with the target cell population.
  • the target cell population can be obtained by removing the cell population of interest from the candidate target cell population.
  • the embodiment of the present application utilizes multi-dimensional parameters for cell analysis, and fully exploits the advantages of the computer in multi-parameter analysis.
  • the embodiments of the present application fully take into account the actual clinical significance of each parameter in the detection item, and use the purpose and function of the fluorescent label addition corresponding to each parameter to break through the cells from a large group (such as lymphocytes) to a subgroup (such as lymphatic sub-population).
  • Group) analysis method which first identifies subgroups or interfering cells in a large group, and then subgroups or interfering cells to assist in determining the analysis mode of the large group, thereby using the reverse gated idea to assist in determining the target cells.
  • the location and distribution edge of the group more accurately determine the location of the target cell population, and distinguish the interfering cells from the target cell population, thereby improving the accuracy of cell sorting, especially for overlapping cell populations or for determining the target cell population. The effect is more obvious.
  • the data processing device 40 includes a streaming multi-dimensional data automatic classification device. As shown in FIG. 5, the device may include: a data acquisition unit 420, an auxiliary parameter determination unit 421, an auxiliary parameter statistics unit 422, and a first extraction. Unit 423, main parameter statistics unit 424, mapping unit 425, and second extraction unit 426.
  • the data acquisition unit 420 is configured to acquire particle feature data for characterizing the cell particles, the particle feature data being a collection of data collected by a plurality of channels of the flow cytometer.
  • the auxiliary parameter determining unit 421 is configured to determine at least one auxiliary parameter according to the detected item, wherein each of the auxiliary parameters is one-dimensional in the particle characteristic data.
  • the auxiliary parameter statistics unit 422 is used to perform statistics on the particle feature data based on the auxiliary parameters.
  • the first extraction unit 423 is for extracting a population of cells of interest from the statistical results of the auxiliary parameters.
  • the main parameter statistics unit 424 is configured to perform statistics on the particle feature data based on the main parameter, where the main parameter is a parameter for circled the target cell group in the statistical result based on the parameter, which is the particle feature data. Different from the one-dimensional of the auxiliary parameters. Mapping unit 425 is used to map the extracted population of cells of interest into the statistical results of the primary parameter. The second extraction unit 426 is configured to utilize the distribution position and edge of the cell population of interest and combine the main parameters to set the target cell population.
  • the auxiliary parameter determining unit 421 obtains the table by looking up the table based on the detected item when determining the auxiliary parameter.
  • the first extraction unit 423 extracts the population of interest cells from the statistical results of the auxiliary parameters based on the characteristics of the test item and the target cell population or the interfering cell population on the auxiliary parameter.
  • the first extraction unit 423 comprises: a cell population dividing subunit 4230 and a cell population determining subunit 4231 of interest.
  • the cell community partitioning subunit 4230 is used to divide the cell population in the statistical results of the auxiliary parameters.
  • cell colony subunit 4230 is used for particle characteristics
  • the data is subjected to threshold processing based on the statistical graph of the auxiliary parameters, and the image after the threshold processing is marked with the communication area, and the cells labeled as one connected area are regarded as one cell community.
  • the cell community dividing sub-unit 4230 is further configured to determine the center of each of the connected areas, and the auxiliary parameter value at the center of the connected area is used as an auxiliary parameter value of the cell community.
  • the cell population determining subunit 4231 of interest is used to determine the cell population in which the auxiliary parameter values are maximal, minimum, or within the set range as the cell population of interest.
  • the cell population of interest belongs to a part of the target cell population
  • the statistical result of the main parameter is a scatter plot
  • the second extracting unit 426 obtains the target cell population by using the distribution position and the edge of the cell population of interest.
  • the distribution area of the cell population of interest is used as the foreground
  • the area outside the set area in the foreground in the scatter plot is used as the background
  • the foreground and the background are segmented, and the boundary between the foreground and the background is found, and the boundary is within the boundary.
  • the part is the target cell group separately.
  • the second extraction unit 426 performs a polygon approximation on the boundary after finding the boundary between the foreground and the background to obtain a polygonal gate, and the cells in the gate are used as the target cell population.
  • peripheral blood lymphocyte subset detection program a case where lymphocytes are gated is further illustrated.
  • Lymphocyte subsets are important indicators for detecting immune function in the body. They are mainly used for the diagnosis and clinical treatment of immune system diseases and immune-related diseases. Common single-label antibodies for lymphocyte subset detection include CD45, CD3, CD4, CD8, CD19, CD16, CD56. Therefore, in the lymphocyte subset detection project, the detected data usually includes forward scattered light, side scattered light, and CD45. , CD3, CD4, CD8, CD19, CD16, CD56 data from multiple fluorescent channels.
  • CD45 is expressed in all leukocytes; CD3 is expressed in T lymphocytes; CD4 is expressed in T helper/inducing lymphocytes (CD4+ T cells) and monocytes; CD8 is expressed in cytotoxic T cells (CD8+ T cells) and NK cells; CD19 is expressed on B lymphocytes; CD16 is expressed in NK cells, monocyte macrophages, granulocytes and dendritic cells; CD56 is expressed in NK fine and fine T-small packets.
  • CD45 is used as a gated antibody to identify lymphocytes, and then lymphocytes are classified on the basis of lymphocytes using specific expression of CD3, CD4, CD8, CD19, CD16, and CD56 in various lymphatic subpopulations. .
  • the main parameters of lymphocyte detection are SSC and CD45.
  • Figure 6a shows the SSC/CD45 scatter plot.
  • the polygon is the gate, and the polygon circle is the lymphocyte.
  • abnormal lymphoid, immature cells, nucleated red blood cells, basophils, and monocytes exist in large numbers or are close to lymphocyte populations, they may interfere with lymphatic gates, or changes in instrument settings such as voltage and compensation.
  • the concentration of the antibody in the reagent changes, or due to an abnormality in the blood sample, or due to a mistake in the sample preparation operation, the position of each group of cells in the scattergram may deviate from the expected position.
  • the lymphocytes determined by only the main parameters SSC and CD45 may be inaccurate, so that the lymphocyte subset analysis results based on the lymphocytes are performed. Not accurate.
  • the present application notifies the reference significance of antibodies such as CD3, CD4, CD8, CD19, CD16, CD56 for recognizing lymphocytes, but they are not specifically labeled with lymphocytes, so it is not possible to directly utilize these antibodies against lymphocytes.
  • Perform cluster analysis Therefore, in the embodiment of the present application, according to the characteristics of the fluorescence parameters of the labeled cells, the cell population is intercepted under other more easily identifiable fluorescence parameters, and then the cell population is found on the gate target scatter plot according to the intercepted cell population. Further analysis. Specifically, the following steps are included:
  • CD3 and CD19 have this characteristic, they are strongly positive in their respective fluorescences, and are separated far from adjacent cells, so CD3 and CD19 are optional parameters.
  • CD3 is expressed in T lymphocytes and CD19 is expressed in B lymphocytes.
  • Figure 6b and Figure 6c are statistical results based on the joint statistics of the auxiliary parameters and other parameters.
  • Figure 6b is a statistical result chart of CD3 combined with SSC
  • Figure 6c is a statistical result chart of CD19 combined with SSC.
  • the region labeled R1 in the figure is a subpopulation of T lymphocytes, which is a group of cells of interest.
  • the region labeled R2 in the figure is the B lymphocyte subset, which is the cell population of interest.
  • the smoothing method can be linear or non-linear smoothing filters such as Gaussian smoothing, mean filtering, and median filtering.
  • the joint area mark is made in Fig. 6d, and the center of each joint area is obtained, as shown by the "*" in Fig. 6e.
  • the blob analysis method is used to extract the center of the connected area, which is equivalent to extracting the position information. Similarly, the size, shape, direction, and quantity of each connected area can be detected.
  • the particle feature data is calculated based on the main parameters and the cell population of interest is mapped to statistical results based on the main parameters for statistics.
  • peripheral blood lymphocyte subsets were detected, using SSC and CD45 as the main parameters.
  • the extracted cell populations of interest R1 and R2 are mapped to statistical results based on the main parameters SSC and CD45, respectively, as shown in Fig. 6g and Fig. 6h.
  • the cell population of interest is only a part of the lymphocytes, but mapping the cell population of interest into the SSC/CD45 scatter plot can indicate the location and edge of the lymphocytes.
  • the target cell population is obtained.
  • a watershed algorithm is used.
  • the population of cells of interest is an identified region, and the distribution of these regions of interest is internally labeled as foreground, as shown in Figure 6i.
  • the mark of the circumscribed rectangle from the inner marked area in FIG. 6i is greater than the r position as the background, where r is a preset value, which can be understood by those skilled in the art, and is not necessarily a circumscribed rectangle, but also other shapes of geometry.
  • Graphics As shown in the horizontal line texture area in Fig. 6j, the watershed algorithm can be used to find the boundary between the foreground and the background, and the part within the boundary is used as the target cell group distribution area, as shown in Fig. 6j, the area surrounded by the curve R3, which is the target cell group. region.
  • the watershed algorithm is used to segment the foreground and background.
  • the foreground and background can be segmented using the active contour algorithm or the random walk algorithm.
  • the polygon approximation can be performed on the region R3 of 6j to obtain the polygon gate, that is, the polygon gate in 6k (the circled portion in the figure).
  • auxiliary parameter is used in combination with other parameters, such as the SSC used in the above embodiment; those skilled in the art should understand that the auxiliary parameter can also be used alone, such as processing the histogram of CD3 and CD19, respectively.
  • Figure 7, II is the extracted population of cells of interest.
  • the auxiliary parameters may also be separately processed.
  • the SSC/CD3, SSC/CD19 scatter plots are processed separately, or the auxiliary parameters are processed jointly, such as processing the CD3/CD19 scatter plot, as shown in Fig. 8, Rl in the figure.
  • the cell population circled by R2 is the extracted cell population of interest.
  • Examples of such automated classification using multidimensional data include the application of a combination of two, three, four, and six color antibodies to a subpopulation of lymphocytes.
  • lymphatic portals are set on the FSC/SSC scatter plot, FSC and SSC are gated parameters, and the lymphocyte population and other surrounding cell populations are located close to each other.
  • CD 14 and CD45 can be used as auxiliary parameters to assist the gate setting.
  • CD45 is strongly positive and CD14 is negative.
  • the upper left part of the scatter plot is That is, R1 circle
  • the region is identified as the helper cell population of interest, mapping the helper cell population of interest to
  • the door can be further automatically set in the SSC/FSC, as in the P1 gate in Figure 9b.
  • nucleated cells need to be classified on the CD45/SSC scattergram.
  • the main gate parameters are SSC and CD45.
  • the naive cells of acute B lymphocytic leukemia patients often appear in nucleated cells.
  • the location of red blood cells, CD19, CD34, and CD10 can be used as auxiliary gate parameters. Cells with positive CD19, CD34, and CD10 parameters are mapped to CD45/SSC scatter plots to determine the location of naive cells. Use a corresponding algorithm to circle the naive cell population.
  • Plasma cells need to be gated in CD45/SSC scatter plots.
  • Plasma cells may be close to or overlap with nucleated red blood cells or naive cells.
  • CD38 or CD138
  • the cells strongly expressed by CD38 (or CD138) are plasma cells, and these plasma cells are mapped onto the CD45/SSC scattergram, and the corresponding algorithm is used to circle the plasma cell population, and the plasma cell population contains expression.
  • Auxiliary parameters can be used not only to identify target cells, but also to exclude interfering cells.
  • the helper cell population extracted from the statistical results of the auxiliary parameters must not be in the target cell population.
  • the target cell population is found in the main set parameters by other methods, and the helper cell population is also mapped.

Abstract

本申请提供的流式分析仪及其多维数据自动分类方法、装置,选定每个检测项目的目标细胞群的辅助参数和主参数,先对细胞的粒子特征数据基于辅助参数进行统计,得到感兴趣细胞群,再对粒子特征数据基于主参数进行统计,然后将感兴趣细胞群映射到基于主参数进行统计的统计结果中,最后利用感兴趣细胞群的分布位置和边缘,结合主参数设门,得到目标细胞群。从而可以提升设门所得的目标细胞群的准确性,进而提升细胞分析的准确性。

Description

一种流式细胞分析仪及其多维数据分类方法、 装置 技术领域
本申请涉及细胞分析领域, 具体涉及一种流式细胞分析仪及其多维 数据自动分类方法、 装置。 背景技术
流式细胞分析仪通过接收激光照射后液流内细胞的各种光信号对细 胞进行分类分析和识别, 流式光信号通常包括前散射光 (FSC ) 、 侧散 射光 (SSC ) 、 各种荧光 (FL1、 FL2 ) , 这些信号构成了流式数 据的不同参数, 或者称为通道, 或者称为维度。 这些光信号可反映细胞 或者微球的物理化学特征, 如大小, 颗粒度和标记的荧光素情况等。
流式细胞分析仪收集各通道的光信号, 利用设门技术分析细胞, 设 门需要在某些维度中指定某一范围的目标细胞群, 对其进行分析。 手动 设门是根据人的主观判断, 不同的人做出的结果具有一定的差异, 难以 达到结果的一致性。 计算机技术为流式细胞术的数据分析提供了便利, 对于许多临床上的流式检测项目, 商业厂商提供了自动设门的功能, 该 功能的重要优势在于不仅可以降低人的工作量, 而且可以减少人工手动 设门的主观判断过程引起的误差, 提高分析结果的一致性。 自动设门的 另一个优势是可以同时分析多个参数, 从而获得更多信息, 有效提升设 门的准确性。
但无论是自动设门还是手动设门来说, 一个共同的难点是: 对于细 胞群分布相互重叠或不易确定目的细胞群的情况,难以得到准确的判断。 例如, 在划分一定范围的细胞群时, 如果在划分的细胞群内存在大量干 扰细胞, 或者干扰细胞距离目标细胞群较近, 都会干扰到设门。 另一方 面, 由于电压或补偿等仪器设置的改变,或者试剂中抗体浓度发生变化, 或者由于血液样本的异常, 或者由于样本制备操作中的失误, 散点图中 各群细胞的位置都可能偏离预期位置。 发明内容
依据本申请的第一方面, 本申请提供一种流式多维数据自动分类方 法, 包括: 获取用于表征细胞粒子的粒子特征数据, 所述粒子特征数据 为通过流式细胞分析仪的多个通道收集的数据集合; 根据检测项目确定 至少一个辅助参数, 所述每个辅助参数为粒子特征数据中的一维; 对粒 子特征数据基于辅助参数进行统计; 从辅助参数的统计结果中提取感兴 趣细胞群; 对粒子特征数据基于主参数进行统计, 所述主参数为最终通 过设门的方式在基于该参数的统计结果中圈出目标细胞群的参数, 其为 粒子特征数据中不同于辅助参数的维度; 将提取的感兴趣细胞群映射到 主参数的统计结果中; 利用感兴趣细胞群的分布位置和边缘, 并结合主 参数设门, 求出目标细胞群。
依据本申请的第二方面, 本申请提供一种流式多维数据自动分类装 置, 包括: 数据获取单元, 用于获取用于表征细胞粒子的粒子特征数据, 所述粒子特征数据为通过流式细胞分析仪的多个通道收集的数据集合; 辅助参数确定单元, 用于根据检测项目确定至少一个辅助参数, 所述每 个辅助参数为粒子特征数据中的一维; 辅助参数统计单元, 用于对粒子 特征数据基于辅助参数进行统计; 第一提取单元, 用于从辅助参数的统 计结果中提取感兴趣细胞群; 主参数统计单元, 用于对粒子特征数据基 于主参数进行统计, 所述主参数为最终通过设门的方式在基于该参数的 统计结果中圈出目标细胞群的参数, 其为粒子特征数据中不同于辅助参 数的维度; 映射单元, 用于将提取的感兴趣细胞群映射到主参数的统计 结果中; 第二提取单元, 用于利用感兴趣细胞群的分布位置和边缘, 并 结合主参数设门, 求出目标细胞群。
依据本申请的第三方面, 本申请提供一种流式细胞分析仪, 包括: 光学检测设备, 用于对被测样本进行光照射, 收集粒子因光照射所产生 的光信息, 并输出与粒子光信息对应的粒子特征数据; 数据处理装置, 用于接收粒子特征数据, 对粒子特征数据进行处理, 所述处理设备包括 上述的流式多维数据自动分类装置。 附图说明
图 1是流式细胞分析仪的原理示意图;
图 2是对粒子的特征数据进行处理的一种流程图;
图 3是提取感兴趣细胞群的流程图;
图 4是利用感兴趣细胞群的分布位置和边缘求出目标细胞群的流程 图;
图 5是一种流式多维数据自动分类装置的结构示意图;
图 6a-图 6k是一种多维数据分类方法的各处理结果图;
图 7是一种提取感兴趣细胞群的处理结果图;
图 8是一种提取感兴趣细胞群的处理结果图;
图 9a是一种提取感兴趣细胞群的处理结果图;
图 9b是一种基于多维数据自动分类方法的设门的处理结果图。 具体实施方式
本申请实施例提供了一种流式细胞分析仪, 请参考图 1 , 为流式细 胞分析仪的原理示意图, 流式细胞分析仪包括光学检测设备 20、 输送设 备 30和数据处理装置 40。
输送设备 30用于将样本液输送到光学检测设备 20中。输送设备 30 通常包括输送管路和控制阃, 样本液通过输送管路和控制阃输送到光学 检测设备 20中。
光学检测设备 20用于对流经其检测区域的样本液进行光照射,通过 多个通道收集细胞 (细胞是非常小的颗粒, 因此细胞也称为粒子) 因光 照射所产生的各种光信息 (例如散射光信息和 /或荧光信息), 并转换成 对应的电信号, 这些信息与粒子的特征对应, 成为粒子特征数据, 即每 个细胞粒子由多个维度的参数组成的集合表征, 该数据的集合可表示成 一个数组, 例如粒子 A由数组 A ( Al,A2,...,Ai )表征。 具体地, 光学检 测设备 20可包括光源 1025、 作为检测区域的流动室 1022、 设置在光轴 上和 /或光轴侧边的光收集装置 1023和光电感应器 1024。 样本液在鞘液 的裹挟下通过提供检测区域的流动室 1022 , 光源 1025发射的光束照射 到检测区域 1021 , 样本液中的各细胞粒子经光束照射后发出散射光(或 者散射光和荧光), 光收集装置 1023对散射光 (或者散射光和荧光) 进 行收集整形, 经收集整形后的光照射到光电感应器 1024 , 光电感应器
1024将光信号转换成对应的电信号输出。
数据处理装置 40用于对接收的粒子的特征数据进行分析处理。 请参考图 2 , 图 2为对粒子的特征数据进行处理的一种流程图, 包 括以下步骤:
步骤 101、 获取粒子特征数据。
其中, 粒子特征数据用于表征细胞粒子。 粒子特征数据为通过流式 细胞分析仪的多个通道收集的数据集合。
步骤 102、 确定辅助参数。
辅助参数是相对主参数定义的, 主参数是指最终用设门的方法圈出 目标细胞群的参数, 主参数通常根据检测项目确定, 即对粒子特征数据 基于选择的参数进行统计, 例如生成直方图或散点图, 在统计结果中通 过设门的方法确定出目标细胞群, 该参数成为该目标细胞群的主参数。 辅助参数是指能够帮助主参数定位目标细胞群或者区分干扰细胞群的参 数。 辅助参数可根据该检测项目所釆用的抗体和经验选择, 例如, 可预 先确定检测项目与辅助参数的对照表, 一个优选的实施例中, 在确定辅 助参数时, 可以根据检测项目, 通过查表获得。
为突出辅助参数的作用, 选择目标细胞或干扰细胞有特异性表达的 参数作为辅助参数, 例如目标细胞或干扰细胞在该参数项的参数值与其 它细胞在该参数项的参数值有明显区别或具有明显的特点。
每个主参数是粒子特征数据中的一维, 每个辅助参数是粒子特征数 据中不同于主参数的一维。
辅助参数可以是一个, 也可以是多个, 具体可以根据检测项目确定 辅助参数。
步骤 103、 对粒子特征数据基于辅助参数进行统计。
一个实施例中, 对粒子特征数据基于辅助参数进行统计可以基于单 个辅助参数进行统计, 例如, 辅助参数为 Al,A2,...,Ai中第 n维数据, 则 对所有细胞粒子第 n维数据 A ( An )进行统计, 形成一维的统计图, 例 如直方图。 在另一个实施例中, 对粒子特征数据基于辅助参数进行统计 可以基于辅助参数和其它参数的联合进行统计, 或者, 基于多个辅助参 数的联合进行统计。 例如对第 n维数据和联合使用的第 1维数据进行统 计, 则所有细胞粒子第 1维和第 n维数据 A ( Al , An )进行统计, 形成 二维的统计图, 例如散点图。
步骤 104、 从辅助参数的统计结果中提取感兴趣细胞群。
其中, 可以根据检测项目和目标细胞群或干扰细胞群在该辅助参数 上的特异性从辅助参数的统计结果中提取感兴趣细胞群。 感兴趣细胞群 用于辅助定位目标细胞群的位置和边缘, 感兴趣细胞群可以是最终的目 标细胞群或其一部分,也可以是干扰细胞。 由于在选择辅助参数时已经 考虑到感兴趣细胞群在辅助参数上的特异性, 因此可以先在辅助参数的 统计结果中划分出细胞群落, 然后根据检测项目和感兴趣细胞群在辅助 参数上的分布特点将符合该特异性的细胞群确定为感兴趣细胞群,例如, 将辅助参数值最大、 最小或处于设定范围内的细胞群落确定为感兴趣细 胞群。
一个优选的实施例中, 提取感兴趣细胞群的流程如图 3所示, 包括 以下步骤: 步骤 1041 , 对粒子特征数据基于辅助参数的统计图进行阔值 处理, 阔值处理的作用是去除图像中灰度值小于阔值的点, 以便去除干 扰, 还可通过阔值处理后将图像变为二值图, 以方便后续处理。
步骤 1042 , 对阈值处理后的图像进行联通区域标记, 将标记为一个 联通区域的细胞作为一个细胞群落。
步骤 1043 , 求出各联通区域的中心, 将该联通区域的中心处的辅助 参数值作为该细胞群落的辅助参数值。
步骤 1044, 确定感兴趣细胞群。 根据感兴趣细胞群在辅助参数上的 分布特点和各细胞群落的辅助参数值, 将符合特异性表达的细胞群落确 定为感兴趣细胞群。
步骤 105、 对粒子特征数据基于主参数进行统计。 同样, 在对所有细胞粒子基于主参数进行统计时, 可以基于单个主 参数进行统计, 形成直方图, 也可以基于主参数和其它参数的联合进行 统计, 形成二维或更多维的散点图。
步骤 106、 将感兴趣细胞群映射到主参数的统计结果中。 在主参数 的统计结果中, 将属于感兴趣细胞群的细胞粒子标记出来。 当有多个感 兴趣细胞群时, 将每个感兴趣细胞群——映射到主参数的统计结果中。
步骤 107、 求目标细胞群。 利用感兴趣细胞群的分布位置和边缘, 并结合主参数设门, 求出目标细胞群。
可釆用分水岭算法、 聚类算法、 等高线法和 /或梯度法找到目标细胞 群和其它细胞的边界, 从而通过设门的方法得到目标细胞群。 一个优选 的实施例中, 如图 4所示, 主参数的统计结果为散点图, 感兴趣细胞群 属于目标细胞群中的一部分, 利用感兴趣细胞群的分布位置和边缘求出 目标细胞群包括以下步骤:
1071、 将感兴趣细胞群的分布区域作为前景。
1072、 将前景外围设定区域以外的区域作为背景。
1073、 对前景和背景进行区域分割, 找到前景和背景之间的边界, 将边界以内的部分作为目标细胞群分布区域。 其中, 对前景和背景进行 区域分割的方法包括分水冷算法、 active contour算法或 random walk算 法。
一个优选的实施例中,在找到前景和背景之间的边界后还可以包括: 对边界进行多边形近似处理, 得到多边形门, 将门内的细胞作为目标细 胞群。
本实施例中, 上述步骤 105也可以先于辅助参数进行统计或与辅助 参数同步进行。
本申请实施例中, 选定每个检测项目的目标细胞群的辅助参数和主 参数, 分别将细胞的粒子特征数据基于辅助参数和主参数进行统计, 从 辅助参数的统计结果中得到感兴趣细胞群, 然后将感兴趣细胞群映射到 基于主参数的统计结果中, 最后利用感兴趣细胞群的分布位置和边缘, 结合主参数设门, 得到目标细胞群。
根据选择的辅助参数和感兴趣细胞群, 一种情况是: 感兴趣细胞群 属于目标细胞群中的一部分, 感兴趣细胞群在主参数统计结果中的分布 情况对目标细胞群的定位和边缘确定具有参考价值, 如上述实施例, 可 根据感兴趣细胞群的分布位置和边缘确定目标细胞群的分布位置和边 缘。 另一种情况是: 感兴趣细胞群是对目标细胞群造成干扰的细胞, 这 种情况下, 当在主参数统计结果中利用设门的方式求出候选目标细胞群 后, 根据感兴趣细胞群在主参数统计结果中的分布情况, 从候选目标细 胞群中剔除感兴趣细胞群即可得到目标细胞群。 本申请实施例一方面利用多维参数进行细胞分析, 充分发挥了计算 机在多参数分析上的优势。 另一方面, 本申请实施例充分注意到了检测 项目中各参数的实际临床意义, 利用各参数对应的荧光标记添加的目的 和功能, 突破细胞从大群 (例如淋巴细胞) 到亚群 (例如淋巴亚群) 的 分析方式, 而釆用先识别出大群中的亚群或干扰细胞, 再由亚群或干扰 细胞来辅助确定大群的分析方式, 从而釆用反向设门的思路辅助确定出 目标细胞群的位置和分布边缘, 更准确确定出目标细胞群的位置, 并将 干扰细胞与目标细胞群区分开来, 提高细胞分类的准确性, 尤其对细胞 群分布相互重叠或不易确定目的细胞群的情况下效果更明显。
基于上述方法,数据处理装置 40包括一种流式多维数据自动分类装 置, 如图 5所示, 该装置可以包括: 数据获取单元 420、 辅助参数确定 单元 421、 辅助参数统计单元 422、 第一提取单元 423、 主参数统计单元 424、 映射单元 425和第二提取单元 426。
数据获取单元 420用于获取用于表征细胞粒子的粒子特征数据, 所 述粒子特征数据为通过流式细胞分析仪的多个通道收集的数据集合。 辅 助参数确定单元 421用于根据检测项目确定至少一个辅助参数, 所述每 个辅助参数为粒子特征数据中的一维。 辅助参数统计单元 422用于对粒 子特征数据基于辅助参数进行统计。 第一提取单元 423用于从辅助参数 的统计结果中提取感兴趣细胞群。 主参数统计单元 424用于对粒子特征 数据基于主参数进行统计, 所述主参数为最终通过设门的方式在基于该 参数的统计结果中圈出目标细胞群的参数, 其为粒子特征数据中不同于 辅助参数的一维。 映射单元 425用于将提取的感兴趣细胞群映射到主参 数的统计结果中。 第二提取单元 426用于利用感兴趣细胞群的分布位置 和边缘, 并结合主参数设门, 求出目标细胞群。
对粒子特征数据基于辅助参数进行统计包括以下任一种:
基于单个辅助参数进行统计。
基于辅助参数和其它参数的联合进行统计。
基于多个辅助参数的联合进行统计。
一个优选的实施例中, 辅助参数确定单元 421在确定辅助参数时根 据检测项目通过查表获得。
一个优选的实施例中, 第一提取单元 423根据检测项目和目标细胞 群或干扰细胞群在该辅助参数上的特点从辅助参数的统计结果中提取感 兴趣细胞群。
一个优选的实施例中, 第一提取单元 423包括: 细胞群落划分子单 元 4230和感兴趣细胞群确定子单元 4231。
细胞群落划分子单元 4230 用于在辅助参数的统计结果中划分细胞 群落。一个优选的实施例中, 细胞群落划分子单元 4230用于对粒子特征 数据基于辅助参数的统计图进行阈值处理, 对阈值处理后的图像进行联 通区域标记, 将标记为一个联通区域的细胞作为一个细胞群落。 细胞群 落划分子单元 4230还用于求出各联通区域的中心 ,将该联通区域的中心 处的辅助参数值作为该细胞群落的辅助参数值。 感兴趣细胞群确定子单 元 4231用于将辅助参数值最大、最小或处于设定范围内的细胞群落确定 为感兴趣细胞群。
一个优选的实施例中, 感兴趣细胞群属于目标细胞群中的一部分, 主参数的统计结果为散点图, 第二提取单元 426在利用感兴趣细胞群的 分布位置和边缘求出目标细胞群时将感兴趣细胞群的分布区域作为前 景, 将散点图中距前景中设定区域的以外的区域作为背景, 对前景和背 景进行区域分割, 找到前景和背景之间的边界, 将边界以内的部分作为 目标细胞群分别区域。 在进一步优选的实施例中, 第二提取单元 426在 找到前景和背景之间的边界后还对边界进行多边形近似处理, 得到多边 形门, 将门内的细胞作为目标细胞群。
下面以外周血淋巴细胞亚群检测项目中通过设门圈出淋巴细胞为例 进一步说明。
淋巴细胞亚群是检测机体免疫功能的重要指标, 临床主要用于免疫 系统疾病及免疫相关疾病的诊断和临床治疗。 用于淋巴细胞亚群检测的 常见单标抗体包括 CD45、 CD3、 CD4、 CD8、 CD19、 CD16、 CD56 , 因 此淋巴细胞亚群检测项目中, 检测的数据通常包括前散射光、 侧散射光 和 CD45、 CD3、 CD4、 CD8、 CD19、 CD16、 CD56 多个荧光通道的数 据。 CD45表达于所有白细胞; CD3表达于 T淋巴细胞; CD4表达于 T 辅助 /诱导淋巴细胞(CD4+T细胞) 和单核细胞; CD8表达于细胞毒 T 细胞 ( CD8+T细胞) 和 NK细胞; CD19表达于 B淋巴细胞; CD16表 达于 NK细胞、 单核巨噬细胞、 粒细胞和树突状细胞等; CD56表达于 NK细 和细 毒 T细月包。
通常釆用 CD45作为设门抗体, 先识别出淋巴细胞, 然后在淋巴细 胞的基础上, 利用 CD3、 CD4、 CD8、 CD19、 CD16、 CD56在各淋巴亚 群的特异性表达, 对淋巴细胞进行分类。
淋巴细胞检测的设门主参数为 SSC、 CD45 , 图 6a 为 SSC/CD45 散点图, 多边形为设门, 多边形圈出的部分为淋巴细胞。 但当异常的淋 巴、 幼稚细胞、 有核红细胞、 嗜碱粒细胞以及单核细胞大量存在或者距 离淋巴细胞群较近时, 都可能干扰到淋巴设门, 或者由于电压、 补偿等 仪器设置的改变, 或者试剂中抗体浓度发生变化, 或者由于血液样本的 异常, 或者由于样本制备操作中的失误, 散点图中各群细胞的位置可能 偏离预期位置。 此时仅依靠设门主参数 SSC、 CD45 确定的淋巴细胞可 能不准确, 从而使得在此淋巴细胞基础上进行的淋巴细胞亚群分析结果 也不准确。在这种情况下,本申请注意到 CD3、 CD4、 CD8、 CD19、 CD16、 CD56 等抗体对识别淋巴细胞的参考意义, 但它们不是特异性标记淋巴 细胞的, 所以不能够直接利用这些抗体对淋巴进行聚类分析。 因此, 本 申请实施例中根据被标记细胞的荧光参数的特点, 在其他更易识别的荧 光参数下截取细胞群, 然后根据截取的细胞群, 在设门目标散点图上找 到该细胞群, 做进一步分析。 具体包括以下步骤:
51、 确定辅助参数。
辅助设门参数可以较好地分离目标细胞群和干扰细胞群。 本例中 CD3和 CD19具有这种特征, 它们在各自的荧光中都是强阳性, 和相邻 细胞分离较远, 所以可选 CD3和 CD19为辅助参数。 CD3表达于 T淋巴 细胞, CD19表达于 B淋巴细胞。
52、 对粒子特征数据基于辅助参数进行统计并提取感兴趣细胞群。 图 6b和图 6c是基于辅助参数和其它参数联合进行统计的统计结果 图。 图 6b为 CD3联合 SSC的统计结果图, 图 6c是 CD19联合 SSC的 统计结果图。
如图 6b所示, 图中标注为 R1的区域是 T淋巴细胞亚群, 即为感兴 趣细胞群。 如图 6c所示, 图中标注为 R2的区域是 B淋巴细胞亚群即为 感兴趣细胞群。
图 6b的 SSC/CD3散点图中自动提取 R1的方法如下:
1)平滑 SSC/CD3散点图, 平滑方法可以是高斯平滑, 均值滤波, 中 值滤波等线性或者非线' f生的平滑滤波器。
2 ) 对散点图进行阔值处理, 得到图 6d。 阔值处理的作用是去除图 像中灰度值小于阔值的点。 设原图为 ( , , (x, 表示像素点的坐标, 处 理后的图像为 ^^, ,
I(x, y), if I(x, y) > threshold
3 ) 对阔值处理后的图像进行联通区域标记。
对图 6d进行联通区域标记,求出个联通区域的中心,如图 6e中 " *" 所标位置。
本例使用的是 blob分析的方法, 提取联通区域的中心, 相当于提取 位置信息, 类似的还可以利用各联通区域的大小、 形状、 方向、 数量等 信息进行检测。
4 ) 比较各联通区域的中心位置, 选取中心在 CD3方向最大的一个 区域, 如图 6f中区域 R1 , 即所提取的第一个感兴趣细胞群 Rl。 5 ) 对于处理 SSC/CD19也使用同样的方法, 即可得到第二个感兴 趣辅助细胞群, 即图 6c中的 R2标出的区域。
6 )对粒子特征数据基于主参数进行统计并将感兴趣细胞群映射到基 于主参数进行统计的统计结果中。
本例中检测外周血淋巴细胞亚群, 利用 SSC和 CD45作为主参数。 将提取的感兴趣细胞群 R1 和 R2分别映射到基于主参数 SSC和 CD45 统计的统计结果图中, 如图 6g和图 6h所示。
本例中感兴趣细胞群只是淋巴细胞中的一部分, 但是将感兴趣细胞 群映射到 SSC/CD45散点图中, 可以指示淋巴细胞的分布位置和边缘。
7 ) 利用辅助细胞群, 结合主设门参数, 求出目标细胞群。
本例中使用分水岭算法, 在 SSC/CD45散点图中, 感兴趣细胞群是 一个已经确定的区域,将这些感兴趣细胞群的分布区域内部标记为前景, 如图 6i。 将图 6i中距内部标记的区域的外接矩形距离大于 r位置的标记 为背景, 其中 r为预设的值, 本领域技术人员可以理解, 并不一定是外 接矩形, 还可以是其它形状的几何图形。 如图 6j中横线紋理区域, 使用 分水岭算法可以找到前景和背景之间的边界, 将边界以内的部分作为目 标细胞群分布区域, 如图 6j 中曲线包围的区域 R3 , 即为目标细胞群的 区域。
本例中使用分水岭算法对前景和背景进行区域分割, 在另一种实施 方式中, 还可以使用 active contour算法或 random walk算法对前景和背 景进行区域分割。
本例中, 可以对 6j的区域 R3进行多边形近似处理, 即可得到多边 形门, 即 6k中的多边形门 (图中圈出部分) 。
上述实施例详细说明了辅助参数与其他参数联合使用的情况, 如上 述实施例中使用了 SSC; 本领域技术人员应当理解, 辅助参数也可以单 独使用, 如分别处理 CD3和 CD19的直方图, 如图 7所示, II为提取的 感兴趣细胞群。 也可以分别处理各辅助参数, 上述实施例中分别处理了 SSC/CD3、SSC/CD19散点图,或者联合处理辅助参数,如处理 CD3/CD19 散点图, 如图 8所示, 图中 Rl、 R2圈出的细胞群为提取的感兴趣细胞 群。
这种利用多维数据进行自动分类的实例包括淋巴亚群的 2色、 3色、 4色、 6色抗体组合方案的应用。
在 2 色抗体组合方案进行淋巴亚群分析方案中, 淋巴门是在 FSC/SSC散点图上设的, FSC、 SSC是设门主参数, 淋巴细胞群和周围 的其他细胞群位置靠近或者有重叠, 为了使淋巴门更可靠, 可以利用 CD 14和 CD45作为辅助参数, 辅助设门, 淋巴细胞 CD45呈强阳性, CD14呈阴性, 如图 9a所示, 将散点图中左上部分的区域, 即 R1圈出 的区域确定为感兴趣的辅助细胞群, 将感兴趣的辅助细胞群映射到
SSC/FSC散点图上, 并进行标记, 例如釆用不同颜色进行标记, 根据该 标记的散点的分布, 可以进一步在 SSC/FSC 自动设门, 如图 9b中的 P1 门。
白血病免疫表型分析中, 需要在 CD45/SSC散点图上对有核细胞进 行分类, 此时, 主设门参数是 SSC、 CD45 , 比如急性 B淋巴细胞白血 病患者的幼稚细胞经常出现在有核红细胞的位置, CD19、 CD34、 CD10 可作为辅助设门参数, 提取 CD19、 CD34、 CD10三个参数均为阳性的 细胞, 映射到 CD45/SSC散点图上, 可用以确定幼稚细胞的位置, 再使 用相应的算法圈出幼稚细胞群。
对于多发性骨髓瘤患者, 正常骨髓标本中, 需要在 CD45/SSC散点 图对浆细胞进行设门, 浆细胞可能和有核红细胞或者幼稚细胞的位置比 较近或者有重叠, 可用 CD38 (或 CD138 )作为辅助参数, CD38 (或 CD138 ) 强表达的细胞为浆细胞, 将此类浆细胞映射到 CD45/SSC散点 图上, 再使用相应的算法圈出浆细胞群, 浆细胞群内包含表达 CD38 (或 CD138 ) 的浆细胞和不表达 CD38 (或 CD138 ) 的浆细胞。
辅助参数不仅可以用于标出目标细胞, 也可以用于排除干扰细胞。 这种情况下在辅助参数的统计结果中提取出的感兴趣辅助细胞群必定不 在目标细胞群中, 通过其他的方法在主设门参数中找到目标细胞群, 同 样也将感兴趣辅助细胞群映射到主参数的统计结果中, 可使用辅助细胞 群验证找的对不对, 然后做出处理, 比如在目标细胞群中减去要排除的 细胞, 或者输出提示信息请用户审核结果。 本领域技术人员可以理解, 上述实施方式中各种方法的全部或部分 步骤可以通过程序来指令相关硬件完成, 该程序可以存储于一计算机可 读存储介质中, 存储介质可以包括: 只读存储器、 随机存储器、 磁盘或 光盘等。
以上应用了具体个例对本申请进行阐述, 只是用于帮助理解本申请 并不用以限制本申请。对于本领域的一般技术人员,依据本申请的思想, 可以对上述具体实施方式进行变化。

Claims

权 利 要 求
1. 一种流式多维数据自动分类方法, 其特征在于, 包括: 获取用于表征细胞粒子的粒子特征数据, 所述粒子特征数据为通过 流式细胞分析仪的多个通道收集的数据集合;
根据检测项目确定至少一个辅助参数, 所述每个辅助参数为粒子特 征数据中的一维;
对粒子特征数据基于辅助参数进行统计;
从辅助参数的统计结果中提取感兴趣细胞群;
对粒子特征数据基于主参数进行统计, 所述主参数为最终通过设门 的方式在基于该参数的统计结果中圈出目标细胞群的参数, 其为粒子特 征数据中不同于辅助参数的维度;
将提取的感兴趣细胞群映射到主参数的统计结果中;
利用感兴趣细胞群的分布位置和边缘, 并结合主参数设门, 求出目 标细胞群。
2.如权利要求 1所述的方法, 其特征在于: 对粒子特征数据基于辅 助参数进行统计包括以下任一种:
基于单个辅助参数进行统计;
基于辅助参数和其它参数的联合进行统计;
基于多个辅助参数的联合进行统计。
3.如权利要求 1所述的方法, 其特征在于确定辅助参数时根据检测 项目通过查表获得。
4.如权利要求 1-3 中任一项所述的方法, 其特征在于, 根据检测项 目和目标细胞群或干扰细胞群在该辅助参数上的特异性从辅助参数的统 计结果中提取感兴趣细胞群。
5.如权利要求 4所述的方法, 其特征在于, 从辅助参数的统计结果 中提取感兴趣细胞群包括:
在辅助参数的统计结果中划分细胞群落;
将辅助参数值最大、 最小或处于设定范围内的细胞群落确定为感兴 趣细胞群。
6.如权利要求 5所述的方法, 其特征在于, 在辅助参数的统计结果 中划分细胞群落包括:
对粒子特征数据基于辅助参数的统计图进行阈值处理;
对阔值处理后的图像进行联通区域标记;
将标记为一个联通区域的细胞作为一个细胞群落。
7. 如权利要求 6所述的方法, 其特征在于还包括:
求出各联通区域的中心;
将该联通区域的中心处的辅助参数值作为该细胞群落的辅助参数 值。
8.如权利要求 1-7 中任一项所述的方法, 其特征在于: 所述主参数 的统计结果为散点图, 利用感兴趣细胞群的分布位置和边缘求出目标细 胞群包括:
将感兴趣细胞群的分布区域作为前景;
将散点图中距前景设定区域以外的区域作为背景;
对前景和背景进行区域分割, 找到前景和背景之间的边界, 将边界 以内的部分作为目标细月包群分布区域。
9.如权利要求 8所述的方法, 其特征在于, 对前景和背景进行区域 分割的方法包括分水冷算法、 active contour算法或 random walk算法。
10. 如权利要求 8所述的方法, 其特征在于, 找到前景和背景之间 的边界后还包括: 对边界进行多边形近似处理, 得到多边形门, 将门内 的细胞作为目标细胞群。
1 1. 如权利要求 1-7中任一项所述的方法, 其特征在于: 利用感兴 趣细胞群的分布位置和边缘求出目标细胞群的算法包括: 聚类算法、 等 高线法和 /或梯度法。
12.—种流式多维数据自动分类装置, 其特征在于, 包括:
数据获取单元, 用于获取用于表征细胞粒子的粒子特征数据, 所述 粒子特征数据为通过流式细胞分析仪的多个通道收集的数据集合;
辅助参数确定单元, 用于根据检测项目确定至少一个辅助参数, 所 述每个辅助参数为粒子特征数据中的一维;
辅助参数统计单元, 用于对粒子特征数据基于辅助参数进行统计; 第一提取单元, 用于从辅助参数的统计结果中提取感兴趣细胞群; 主参数统计单元, 用于对粒子特征数据基于主参数进行统计, 所述 主参数为最终通过设门的方式在基于该参数的统计结果中圈出目标细胞 群的参数, 其为粒子特征数据中不同于辅助参数的维度;
映射单元,用于将提取的感兴趣细胞群映射到主参数的统计结果中; 第二提取单元, 用于利用感兴趣细胞群的分布位置和边缘, 并结合 主参数设门, 求出目标细胞群。
13. 如权利要求 12所述的装置, 其特征在于: 对粒子特征数据基 于辅助参数进行统计包括以下任一种:
基于单个辅助参数进行统计;
基于辅助参数和其它参数的联合进行统计;
基于多个辅助参数的联合进行统计。
14. 如权利要求 12所述的装置, 其特征在于辅助参数确定单元在 确定辅助参数时根据检测项目通过查表获得。
15. 如权利要求 12-14中任一项所述的装置, 其特征在于, 第一提 取单元根据检测项目和目标细胞群或干扰细胞群在该辅助参数上的特异 性从辅助参数的统计结果中提取感兴趣细胞群。
16. 如权利要求 15所述的装置, 其特征在于, 第一提取单元包括: 细胞群落划分子单元,用于在辅助参数的统计结果中划分细胞群落; 感兴趣细胞群确定子单元, 用于将辅助参数值最大、 最小或处于设 定范围内的细胞群落确定为感兴趣细胞群。
17. 如权利要求 16所述的装置, 其特征在于, 细胞群落划分子单 元用于对粒子特征数据基于辅助参数的统计图进行阈值处理, 对阈值处 理后的图像进行联通区域标记, 将标记为一个联通区域的细胞作为一个 细胞群落。
18. 如权利要求 17所述的装置, 其特征在于细胞群落划分子单元 还用于求出各联通区域的中心, 将该联通区域的中心处的辅助参数值作 为该细胞群落的辅助参数值。
19. 如权利要求 12-18中任一项所述的装置, 其特征在于: 所述主 参数的统计结果为散点图, 第二提取单元在利用感兴趣细胞群的分布位 置和边缘求出目标细胞群时将感兴趣细胞群的分布区域作为前景, 将散 点图中距前景设定区域以外的区域作为背景, 对前景和背景进行区域分 割, 找到前景和背景之间的边界, 将边界以内的部分作为目标细胞群分 布区域。
20. 如权利要求 19所述的装置, 其特征在于第二提取单元在找到 前景和背景之间的边界后还对边界进行多边形近似处理,得到多边形门, 将门内的细胞作为目标细胞群。
21.—种流式细胞分析仪, 其特征在于包括:
光学检测设备, 用于对被测样本进行光照射, 收集粒子因光照射所 产生的光信息, 并输出与粒子光信息对应的粒子特征数据;
数据处理装置,用于接收粒子特征数据,对粒子特征数据进行处理, 所述处理设备包括如权利要求 12-20中任一项所述的流式多维数据自动 分类装置。
PCT/CN2014/075613 2014-04-17 2014-04-17 一种流式细胞分析仪及其多维数据分类方法、装置 WO2015157970A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
CN201480074432.5A CN105940301B (zh) 2014-04-17 2014-04-17 一种流式细胞分析仪及其多维数据分类方法、装置
PCT/CN2014/075613 WO2015157970A1 (zh) 2014-04-17 2014-04-17 一种流式细胞分析仪及其多维数据分类方法、装置
US15/295,891 US20170102310A1 (en) 2014-04-17 2016-10-17 Flow cytometer and a multi-dimensional data classification method and an apparatus thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2014/075613 WO2015157970A1 (zh) 2014-04-17 2014-04-17 一种流式细胞分析仪及其多维数据分类方法、装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/295,891 Continuation US20170102310A1 (en) 2014-04-17 2016-10-17 Flow cytometer and a multi-dimensional data classification method and an apparatus thereof

Publications (1)

Publication Number Publication Date
WO2015157970A1 true WO2015157970A1 (zh) 2015-10-22

Family

ID=54323400

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/075613 WO2015157970A1 (zh) 2014-04-17 2014-04-17 一种流式细胞分析仪及其多维数据分类方法、装置

Country Status (3)

Country Link
US (1) US20170102310A1 (zh)
CN (1) CN105940301B (zh)
WO (1) WO2015157970A1 (zh)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108267571A (zh) * 2017-01-03 2018-07-10 中国医学科学院医学实验动物研究所 一种血液种属判别的方法
CN106706938B (zh) * 2017-02-14 2018-03-09 迈克医疗电子有限公司 一种样本测试方法、样本测试控制装置及样本测试系统
US11573182B2 (en) 2017-05-25 2023-02-07 FlowJo, LLC Visualization, comparative analysis, and automated difference detection for large multi-parameter data sets
CN108375675A (zh) * 2018-01-29 2018-08-07 李小峰 外周血淋巴细胞亚群细胞浓度检测试剂盒及其检测方法
CN113260848A (zh) * 2018-12-21 2021-08-13 Abs全球公司 亚群识别的系统和方法
WO2020146733A1 (en) * 2019-01-11 2020-07-16 Becton, Dickinson And Company Optimized sorting gates
US11977017B2 (en) * 2019-01-23 2024-05-07 International Business Machines Corporation Automated configuration of flow cytometry machines
CN113188982B (zh) * 2021-04-30 2022-05-10 天津深析智能科技发展有限公司 淋巴细胞亚群自动分析中有效去除单核细胞干扰的方法
CN113380318B (zh) * 2021-06-07 2023-04-07 天津金域医学检验实验室有限公司 人工智能辅助流式细胞术40cd免疫表型检测方法及系统
CN114720681A (zh) * 2022-05-11 2022-07-08 深圳市帝迈生物技术有限公司 一种样本分析仪及其多联检滤波的方法
CN116642819B (zh) * 2023-07-19 2023-10-10 江苏得康生物科技有限公司 细胞群的识别方法及其装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101226190A (zh) * 2007-01-17 2008-07-23 深圳迈瑞生物医疗电子股份有限公司 流式细胞术的自动分类方法和装置
CN102227624A (zh) * 2008-11-04 2011-10-26 贝克曼考尔特公司 用于显示三维物体散点图的系统和方法
CN102507417A (zh) * 2011-11-29 2012-06-20 长春迪瑞医疗科技股份有限公司 一种粒子自动分类方法
CN103364324A (zh) * 2012-03-27 2013-10-23 嘉善加斯戴克医疗器械有限公司 一种用于血液细胞分析仪的自适应分类计算方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011515088A (ja) * 2008-03-22 2011-05-19 メルク・シャープ・エンド・ドーム・コーポレイション 増殖因子シグナル伝達経路レギュレーション状態を評価するための方法及び遺伝子発現サイン
CN102305758B (zh) * 2011-05-19 2014-01-08 长春迪瑞医疗科技股份有限公司 一种粒子快速自动分类方法及其实现装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101226190A (zh) * 2007-01-17 2008-07-23 深圳迈瑞生物医疗电子股份有限公司 流式细胞术的自动分类方法和装置
CN102227624A (zh) * 2008-11-04 2011-10-26 贝克曼考尔特公司 用于显示三维物体散点图的系统和方法
CN102507417A (zh) * 2011-11-29 2012-06-20 长春迪瑞医疗科技股份有限公司 一种粒子自动分类方法
CN103364324A (zh) * 2012-03-27 2013-10-23 嘉善加斯戴克医疗器械有限公司 一种用于血液细胞分析仪的自适应分类计算方法

Also Published As

Publication number Publication date
US20170102310A1 (en) 2017-04-13
CN105940301A (zh) 2016-09-14
CN105940301B (zh) 2017-06-27

Similar Documents

Publication Publication Date Title
WO2015157970A1 (zh) 一种流式细胞分析仪及其多维数据分类方法、装置
Patel et al. Automated leukaemia detection using microscopic images
AU2009212193B2 (en) Method and system for analysis of flow cytometry data using support vector machines
US10337975B2 (en) Method and system for characterizing particles using a flow cytometer
US9772282B2 (en) System for wide field-of-view, highly oblique illumination microscopy for scatter-based discrimination of cells
US20180003634A1 (en) Nucleated red blood cell warning method and device, and flow cytometer using the same
US20140221247A1 (en) System, method, and article for detecting abnormal cells using multi-dimensional analysis
US9658215B2 (en) Identifying and enumerating early granulated cells (EGCs)
JP2019512697A (ja) デジタルホログラフィ顕微鏡検査および無傷の(untouched)末梢血白血球を用いる高精度の5部鑑別(5−part Differential)
CN105136795A (zh) 血液样本检测装置、方法和系统
US11644407B2 (en) Adaptive sorting for particle analyzers
CN107389536B (zh) 基于密度-距离中心算法的流式细胞粒子分类计数方法
Rodrigues et al. Optimized automated data analysis for the cytokinesis‐block micronucleus assay using imaging flow cytometry for high throughput radiation biodosimetry
CN105980852A (zh) 细胞分析方法和系统、装置
CN105986003B (zh) 一种白细胞计数方法、装置及细胞分析仪
EP2869062A1 (en) Urine sample analyzer and urine sample analyzing method
Loddo et al. A computer-aided system for differential count from peripheral blood cell images
JP2015096847A (ja) 細胞分析方法、癌化情報提供方法および癌化情報提供装置
Meimban et al. Blood cells counting using python opencv
CN110226083B (zh) 红细胞碎片识别方法和装置、血液细胞分析仪及分析方法
GB2428471A (en) Flow cytometry
Priyankara et al. An extensible computer vision application for blood cell recognition and analysis
CN113380318B (zh) 人工智能辅助流式细胞术40cd免疫表型检测方法及系统
EP3244191A1 (en) Method and system for characterizing particles using a flow cytometer
Norton et al. Cell enumeration and characterization in microvolume laser scanning cytometry: a multicolor image-processing package

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14889508

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 23.12.2016)

122 Ep: pct application non-entry in european phase

Ref document number: 14889508

Country of ref document: EP

Kind code of ref document: A1