[go: up one dir, main page]

WO2021256514A1 - Dispositif, système, programme et procédé d'analyse de cellules vivantes - Google Patents

Dispositif, système, programme et procédé d'analyse de cellules vivantes Download PDF

Info

Publication number
WO2021256514A1
WO2021256514A1 PCT/JP2021/022892 JP2021022892W WO2021256514A1 WO 2021256514 A1 WO2021256514 A1 WO 2021256514A1 JP 2021022892 W JP2021022892 W JP 2021022892W WO 2021256514 A1 WO2021256514 A1 WO 2021256514A1
Authority
WO
WIPO (PCT)
Prior art keywords
living
living cell
cell analysis
unit
cell
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/JP2021/022892
Other languages
English (en)
Japanese (ja)
Inventor
大輝 中矢
聖也 内藤
源志郎 安部
史希 佐藤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
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 Individual filed Critical Individual
Priority to JP2022531881A priority Critical patent/JP7427289B2/ja
Publication of WO2021256514A1 publication Critical patent/WO2021256514A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12M—APPARATUS FOR ENZYMOLOGY OR MICROBIOLOGY; APPARATUS FOR CULTURING MICROORGANISMS FOR PRODUCING BIOMASS, FOR GROWING CELLS OR FOR OBTAINING FERMENTATION OR METABOLIC PRODUCTS, i.e. BIOREACTORS OR FERMENTERS
    • C12M3/00—Tissue, human, animal or plant cell, or virus culture apparatus
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/02—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving viable microorganisms
    • C12Q1/04—Determining presence or kind of microorganism; Use of selective media for testing antibiotics or bacteriocides; Compositions containing a chemical indicator therefor
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/27—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands using photo-electric detection ; circuits for computing concentration
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/483—Physical analysis of biological material
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis

Definitions

  • the present invention relates to a technique for analyzing a living cell using hyperspectral data, and more particularly to a living cell analysis device, a living cell analysis system, a living cell analysis program, and a living cell analysis method that can be applied to cytodiagnosis. ..
  • pathological diagnosis such as histological diagnosis (biopsy) and cytopathology
  • morphological diagnosis using a microscope by a pathologist is performed.
  • a morphological diagnosis depends largely on the knowledge and experience of a pathologist, and is so difficult that it requires several years of training to become a full-fledged person.
  • Patent Document 1 proposes a microscope capable of acquiring an image of a pathological specimen used for cytopathology.
  • the pathological specimen for histology is prepared by slicing the tissue collected from the subject
  • the pathological specimen for cytopathology is prepared without slicing the cells collected from the subject.
  • There is thickness. Therefore, the pathological specimen of cytopathology has a problem that it becomes a bottleneck in diagnostic imaging because it is difficult to focus a microscopic image due to its thickness and it is difficult to discriminate cells.
  • the present invention has been made to solve such a problem, and is a living cell analysis device, a living cell analysis system, which can automatically and accurately analyze living cells contained in an entire pathological specimen. It is an object of the present invention to provide a living cell analysis program and a living cell analysis method.
  • the biological cell analysis apparatus associates spectral information with each pixel constituting a two-dimensional image in order to solve the problem of automatically and accurately analyzing biological cells contained in the entire pathological specimen.
  • It is a living cell analysis device that analyzes living cells using hyperspectral data, and the stage on which the pathological specimen containing the living cells is placed and the hyperspectral camera are relatively moved to move the entire pathological specimen.
  • a hyperspectral data acquisition unit that acquires hyperspectral data of the entire pathological specimen from the hyperspectral camera, and a two-dimensional image of the entire pathological specimen based on the hyperspectral data. It has a living cell discrimination unit that discriminates pixels corresponding to living cells by machine learning.
  • a spectrum information extraction unit for extracting spectral information of pixels corresponding to the living cell. It may have a alteration state identification unit for identifying the alteration state of the living cell by classifying the spectral information of the pixel corresponding to the living cell by machine learning.
  • the altered state may be any of the following (i) to (iv). (i) Whether or not the living cell is a cancer cell; (ii) Which site is the cancer cell; (iii) Cancer progression or malignancy; (iv) Infiltration of ulcers or inflammation.
  • a trained cell discrimination model in which the spectral information of the living cells is pre-learned as teacher data is learned. Is prepared for each part of the body, and the living cell discrimination unit may determine whether or not the cell corresponds to the living cell by the cell discrimination model.
  • the spectral information corresponding to the altered state of the living cell is used as teacher data in advance.
  • a trained trained altered state identification model is prepared for each altered state, and the altered state identification unit may identify the altered state of the living cell by the altered state identification model.
  • the biological cell analysis system collects spectral information for each pixel constituting a two-dimensional image in order to solve the problem of automatically and accurately analyzing biological cells contained in the entire pathological specimen.
  • It is a living cell analysis system that analyzes living cells using the associated hyperspectral data
  • the living cell analysis system is composed of a living cell analysis device and a living cell analysis server capable of data communication with each other.
  • the living cell analysis apparatus has a whole scan instruction unit that relatively moves a stage on which a pathological specimen containing the living cells is placed and a hyperspectral camera to scan the entire pathological specimen, and the hyperspectral camera.
  • a hyperspectral data acquisition unit that acquires hyperspectral data of the entire pathological specimen, and a device-side communication unit that transmits the hyperspectral data to a biological cell analysis server and receives analysis results from the biological cell analysis server.
  • the living cell analysis server has a display control unit for displaying the analysis result on a display means, and the living cell analysis server mechanically obtains pixels corresponding to the living cells from a two-dimensional image of the entire pathological specimen based on the hyperspectral data.
  • a biological cell discrimination unit that discriminates by learning, and a server-side communication unit that receives the hyperspectral data from the biological cell analysis device and transmits analysis results including discrimination results by the biological cell discrimination unit to the biological cell analysis device. And have.
  • the living cell analysis program collects spectral information for each pixel constituting a two-dimensional image in order to solve the problem of automatically and accurately analyzing living cells contained in the entire pathological specimen. It is a living cell analysis program that analyzes living cells using the associated hyperspectral data.
  • the stage on which the pathological specimen containing the living cells is placed and the hyperspectral camera are relatively moved, and the pathological specimen is moved.
  • a whole scan instruction unit that scans the entire path, a hyperspectral data acquisition unit that acquires hyperspectral data of the entire pathological specimen from the hyperspectral camera, and a two-dimensional image of the entire pathological specimen based on the hyperspectral data.
  • the computer is made to function as a living cell discrimination unit that discriminates pixels corresponding to the living cells by machine learning.
  • the living cell analysis method provides spectral information for each pixel constituting a two-dimensional image in order to solve the problem of automatically and accurately analyzing living cells contained in the entire pathological specimen. It is a living cell analysis method for analyzing living cells using the associated hyperspectral data, in which the stage on which the pathological specimen containing the living cells is placed and the hyperspectral camera are relatively moved to obtain the pathological specimen.
  • a two-dimensional image of the entire pathological specimen based on the whole scan instruction step for scanning the entire path, the hyperspectral data acquisition step for acquiring the hyperspectral data of the entire pathological specimen from the hyperspectral camera, and the hyperspectral data. It has a living cell discrimination step for discriminating pixels corresponding to the living cells from the above by machine learning.
  • living cells contained in the entire pathological specimen can be analyzed automatically and with high accuracy.
  • the biological cell analysis system 100A of the first embodiment mainly includes a stage 11 on which a pathological specimen 10 containing a biological cell is placed, a hyperspectral camera 12 for acquiring hyperspectral data, and a hyperspectral camera 12. It is composed of a living cell analysis device 1A that analyzes living cells based on the hyperspectral data acquired by the hyperspectral camera 12.
  • Stage 11 is for placing a pathological specimen 10 containing living cells.
  • the stage 11 is configured to be horizontally movable by a stepping motor (not shown), and the moving direction and the moving amount are adjusted according to the drive signal from the biological cell analysis device 1A. Further, the stage 11 is configured to transmit light from the light source 13 provided below in a state where the pathological specimen 10 such as a slide is set.
  • the stage 11 is moved in order to scan the entire pathological specimen 10 with the hyperspectral camera 12, but the stage 11 and the hyperspectral camera are not limited to this configuration. Any configuration may be used as long as the 12 is relatively moved. That is, the stage 11 may be fixed and the hyperspectral camera 12 may be moved, or both the stage 11 and the hyperspectral camera 12 may be moved.
  • the hyperspectral camera 12 simultaneously acquires two-dimensional spatial information and spectral information (hyperspectral information) at a plurality of wavelengths. Specifically, as shown in FIG. 2, the hyperspectral camera 12 acquires hyperspectral data in which spectral information is associated with each pixel constituting the two-dimensional image.
  • the spectral information is also called a spectral spectrum and shows the distribution of light intensity for each band (wavelength).
  • the hyperspectral camera 12 has a magnifying lens 12a, and the transmitted light transmitted through the pathological specimen 10 is magnified and measured. Visible light has a smaller photon energy than X-rays and has less effect on the human body.
  • Visible light has a smaller photon energy than X-rays and has less effect on the human body.
  • a visible light source is used as the light source 13 so that visible light is included in the wavelength band of the hyperspectral data.
  • the hyperspectral camera 12 may incorporate a lens for correcting aberrations, such as a collimator lens and a telecentric lens.
  • a lens for correcting aberrations such as a collimator lens and a telecentric lens.
  • the light source 13 a halogen lamp, a high color LED (Light Emitting Diode), a xenon lamp, a semiconductor laser light source, or the like can be used.
  • the biological cell analysis device 1A is composed of a computer such as a personal computer or a tablet, and as shown in FIG. 1, mainly a display means 2, an input means 3, a storage means 4, and an arithmetic processing means. Has 5 and.
  • a computer such as a personal computer or a tablet
  • FIG. 1 mainly a display means 2, an input means 3, a storage means 4, and an arithmetic processing means. Has 5 and.
  • each configuration means will be described in detail.
  • the display means 2 is composed of a liquid crystal display or the like, and displays the analysis result or the like by the living cell analysis device 1A.
  • the input means 3 is composed of a keyboard, a mouse, or the like, and inputs instructions and selections by the user.
  • the display means 2 having only the display function and the input means 3 having only the input function are used separately, but the present invention is not limited to this configuration, and the display function is not limited to this.
  • a display input means such as a touch panel having an input function may be used.
  • the storage means 4 stores various data and functions as a working area when the arithmetic processing means 5 performs arithmetic processing.
  • the storage means 4 is composed of a hard disk, a ROM (ReadOnlyMemory), a RAM (RandomAccessMemory), a flash memory, and the like, and as shown in FIG. 1, the program storage unit 41 and It has a cell discrimination parameter storage unit 42 and a alteration state identification parameter storage unit 43.
  • ROM ReadOnlyMemory
  • RAM RandomAccessMemory
  • flash memory and the like, and as shown in FIG. 1, the program storage unit 41 and It has a cell discrimination parameter storage unit 42 and a alteration state identification parameter storage unit 43.
  • the biological cell analysis program 1a for controlling the biological cell analysis device 1A of the first embodiment is installed in the program storage unit 41. Then, the arithmetic processing means 5 executes the biological cell analysis program 1a to make the computer as the biological cell analysis device 1A function as each component described later.
  • the usage pattern of the biological cell analysis program 1a is not limited to the above configuration.
  • the living cell analysis program 1a may be stored in a non-temporary recording medium such as a CD-ROM or a DVD-ROM that can be read by a computer, and may be directly read from the recording medium and executed. Further, it may be used by a cloud computing method or an ASP (Application Service Provider) method from an external server or the like.
  • a cloud computing method or an ASP (Application Service Provider) method from an external server or the like.
  • the cell discrimination parameter storage unit 42 stores the learned cell discrimination parameters obtained by prior learning.
  • a trained cell discrimination model obtained by applying cell discrimination parameters to a learning / inference model which is a machine learning algorithm is adopted. There is. Therefore, the cell discrimination parameter storage unit 42 stores cell discrimination parameters obtained by pre-learning the cell discrimination model using the spectral information of living cells as teacher data.
  • the cell discrimination parameter storage unit 42 stores the cell discrimination parameters applied to each cell discrimination model for each body part.
  • body parts include the large intestine, ovary, breast, pancreas, uterus, prostate, skin, stomach, and lungs, which are prone to develop cancer.
  • the training data to be learned by the cell discrimination model includes immune cells (lymphocytes, etc.) that are the background of living cells in order to improve the discrimination accuracy of living cells. ), Mucilage, interstitium and spectral information such as unphotographed cells are used.
  • the altered state identification parameter storage unit 43 stores the learned altered state identification parameters obtained by prior learning.
  • a learned altered state identification is performed by applying a parameter for identifying the altered state to a learning / inference model which is a machine learning algorithm. The model is adopted. Therefore, the alteration state identification parameter storage unit 43 stores the alteration state identification parameter obtained by pre-learning the alteration state identification model using the spectral information corresponding to the alteration state of the living cell as teacher data.
  • the altered state is a concept including all states indicating whether or not the living cells are altered or to what extent.
  • a trained altered state identification model is prepared for each altered state shown in (i) to (iv) below, and can be selected by the user. It is configured in. (i) Whether or not the living cell is a cancer cell (ii) Which organ is the cancer cell (iii) Progression or malignancy of the cancer (iv) Infiltration of ulcer or inflammation Therefore, for identification of alteration state
  • the parameter storage unit 43 stores the alteration state identification parameters applied to each alteration state identification model for each alteration state.
  • the spectral information of healthy living cells and the spectral information of cancer cells are used as the teacher data to be trained by the altered state discriminative model. Further, in the case of the altered state of (ii) above, the spectral information of cancer cells in each part of the body is used. In the case of (iii) above, spectral information according to the degree of cancer progression and malignancy is used. For example, the degree of progression is specified in four stages: non-cancer cells, mild dysplasia (LGD: Low Grade Dysplasia), high dysplasia (HGD: High Grade Dysplasia), and cancer cells. In the case of (iv) above, spectral information according to the degree of infiltration of ulcer or inflammation is used.
  • the arithmetic processing means 5 executes analysis processing of living cells based on hyperspectral data.
  • the arithmetic processing means 5 is composed of a CPU (Central Processing Unit) or the like, and is as shown in FIG. 1 by executing the biological cell analysis program 1a installed in the storage means 4.
  • Overall scan instruction unit 51 hyperspectral data acquisition unit 52, input reception unit 53, living cell discrimination unit 54, spectrum information extraction unit 55, preprocessing execution unit 56, alteration state identification unit 57, and so on. It functions as a display control unit 58.
  • display control unit 58 As a display control unit 58.
  • the whole scan instruction unit 51 scans the entire pathological specimen 10 with the hyperspectral camera 12.
  • the overall scan instruction unit 51 relatively moves the stage 11 and the hyperspectral camera 12 by outputting a drive signal to the stepping motor that horizontally moves the stage 11, and the pathological specimen 10 It is designed to scan the entire screen.
  • the hyperspectral data acquisition unit 52 acquires hyperspectral data of the entire pathological specimen 10 from the hyperspectral camera 12.
  • the hyperspectral data acquisition unit 52 sequentially acquires the hyperspectral data output from the hyperspectral camera 12 during the scanning operation, and acquires the hyperspectral data of the entire pathological specimen 10. There is.
  • the input receiving unit 53 receives inputs such as selections and instructions from the user.
  • a trained cell discrimination model is prepared for each body part.
  • trained alteration state identification models are prepared for each of the various alteration states. Therefore, when the user selects any part or the altered state via the input means 3, the input receiving unit 53 accepts the selection of the learned model corresponding to the said portion or the altered state.
  • the living cell discrimination unit 54 discriminates living cells contained in the pathological specimen 10.
  • the living cell discrimination unit 54 uses machine learning to obtain pixels corresponding to living cells from a two-dimensional image of the entire pathological specimen 10 based on the hyperspectral data acquired by the hyperspectral data acquisition unit 52. Determine.
  • the living cell discrimination unit 54 is configured by a trained cell discrimination model obtained by applying a cell discrimination parameter to a learning / inference model which is a machine learning algorithm. Therefore, when the living cell discrimination unit 54 discriminates the living cell, the body portion received by the input receiving unit 53 among the learned cell discrimination parameters stored in the cell discrimination parameter storage unit 42. The corresponding cell discrimination parameters are applied to the cell discrimination model. Then, the learned cell discrimination model is used to discriminate and label the pixels corresponding to the living cells from the two-dimensional image of the entire pathological specimen 10.
  • a support vector machine which is one of supervised learning, is used as an algorithm of the living cell discrimination unit 54, but the present invention is not limited to this. Instead, other supervised learning represented by decision tree learning, k-nearest neighbor method (k-Nearest Neighbor: k-NN), random forest, neural network related methods such as deep learning, etc. may be adopted.
  • SVM Support Vector Machine
  • the algorithm of the living cell discrimination unit 54 is not limited to supervised learning, and various types of unsupervised learning represented by clustering, principal component analysis, EM (Expectation-Maximization) algorithm, cosine distance, autoencoder, etc. Learning may be adopted.
  • unsupervised learning the groups determined for each pixel are displayed in different colors, and it is determined by the user whether or not the living cells are extracted.
  • semi-supervised learning that can be learned by utilizing a large amount of data without a correct answer label by using a small amount of data with a correct answer label may be adopted.
  • the spectrum information extraction unit 55 extracts spectrum information of pixels corresponding to living cells.
  • the spectrum information extraction unit 55 extracts spectrum information (light intensity for each band) of each pixel corresponding to the living cell discriminated by the living cell discriminating unit 54.
  • the pre-processing execution unit 56 executes various pre-processing to improve the analysis accuracy and the analysis speed of the spectral information.
  • the pre-processing execution unit 56 executes pre-processing such as normalization, whitening, reflectance calculation, and band selection according to selection and setting by the user.
  • normalization normalizes so that the maximum value of the spectral intensity is 1, and the information on the spectral intensity is lost, but the information on the relative waveform is obtained.
  • Whitening eliminates the correlation of each wavelength intensity and emphasizes characteristic spectral changes.
  • the reflectance calculation calculates the absorption degree of the object based on the wavelength of the light source 13, and the characteristic peculiar to the object that does not depend on the change of the light source 13 is extracted. Band selection reduces the number of data and improves the analysis speed by selecting only an arbitrary band from the spectral information.
  • the pre-processing by the pre-processing execution unit 56 improves the analysis accuracy and the analysis speed by performing the pre-processing before the identification by the alteration state identification unit 57, but it is not always necessary to execute the pre-processing.
  • the alteration state identification unit 57 identifies the alteration state of living cells.
  • the alteration state identification unit 57 identifies the alteration state of the living cell by classifying the spectral information of the pixels corresponding to the living cell extracted by the spectrum information extraction unit 55 by machine learning. ..
  • the alteration state identification unit 57 is configured by a trained alteration state identification model obtained by applying a alteration state identification parameter to a learning / inference model that is a machine learning algorithm. Therefore, when the alteration state identification unit 57 identifies the alteration state of the living cell, the input reception unit 53 accepts the learned alteration state identification parameters stored in the alteration state identification parameter storage unit 43. The parameters for identifying the altered state corresponding to the altered state are applied to the altered state identification discrimination model. Then, the learned spectral state identification model is used to classify the spectral information of the pixels corresponding to the living cells.
  • a support vector machine which is one of supervised learning, is used as an algorithm of the alteration state identification unit 57, but the present invention is not limited to this. Instead, other supervised learning represented by decision tree learning, k-nearest neighbor method (k-Nearest Neighbor: k-NN), random forest, neural network related methods such as deep learning, etc. may be adopted.
  • SVM Support Vector Machine
  • the algorithm of the alteration state identification unit 57 is not limited to supervised learning, and various types of unsupervised learning represented by clustering, principal component analysis, EM (Expectation-Maximization) algorithm, cosine distance, autoencoder, etc. Learning may be adopted. Alternatively, semi-supervised learning that can be learned by utilizing a large amount of data without a correct answer label by using a small amount of data with a correct answer label may be adopted.
  • the display control unit 58 is for controlling the content to be displayed on the display means 2.
  • the display control unit 58 causes the display means 2 to display the graphical user interface of the biological cell analysis program 1a, the selected screen of the trained model, the analysis result, and the like.
  • the analysis result is an image of the discrimination result of the living cell and an image of the discrimination result of the altered state.
  • the living cell analysis device 1A the living cell analysis system 100A, the living cell analysis program 1a, and the living cell analysis method of the first embodiment will be described.
  • the living cell analysis system 100A, the living cell analysis program 1a, and the living cell analysis method of the first embodiment in advance.
  • the pathological specimen 10 (preparate) is set on the stage 11. After that, as shown in FIG. 3, first, the whole scan instruction unit 51 relatively moves the stage 11 and the hyperspectral camera 12 to scan the entire pathological specimen 10 (step S1: whole scan instruction step). .. As a result, the entire pathological specimen 10 is automatically scanned, and the living cells are automatically discriminated and analyzed.
  • the hyperspectral data acquisition unit 52 acquires the hyperspectral data of the entire pathological specimen 10 from the hyperspectral camera 12 (step S2: hyperspectral data acquisition step).
  • the light source 13 a visible light source which has higher transparency than infrared rays and ultraviolet rays and is not easily affected by aberrations is used.
  • visible light has less effect on the human body than X-rays, and its spectral spectrum potentially contains information on many cell tissues. Therefore, hyperspectral data suitable for analyzing the altered state of living cells is acquired.
  • the input receiving unit 53 accepts the selection of the learned model to be used for discriminating the living cell and identifying the altered state (step S3: input receiving step).
  • step S3 input receiving step.
  • a cell discrimination model suitable for the site of the living cell to be discriminated is selected, so that the cell discrimination accuracy is improved.
  • the alteration state identification model suitable for the alteration state to be identified is selected, the identification accuracy of the alteration state is improved.
  • the living cell discrimination unit 54 discriminates the pixels corresponding to the living cells from the two-dimensional image of the entire pathological specimen 10 by machine learning based on the hyperspectral data acquired by the hyperspectral data acquisition unit 52 (step).
  • S4 Living cell discrimination step).
  • the spectral information of the hyperspectral data is resistant to focus shift due to its nature and does not easily affect the discrimination accuracy. Therefore, even in the pathological specimen 10 having a relatively thick thickness, the pixels corresponding to the living cells can be discriminated with high accuracy.
  • the counting of living cells becomes easy and accurate. Therefore, it can also be applied to a diagnosis in which positive / negative is determined by the number and ratio of living cells existing in a certain region. It can also be applied to a diagnosis in which the cytoplasm of a living cell is stained and positive / negative is determined based on the staining concentration and the staining ratio within a certain region.
  • the spectrum information extraction unit 55 extracts spectrum information of pixels corresponding to living cells (step S5: spectrum information extraction step). As a result, the spectral information of the pixels corresponding to the living cells is output as analysis data.
  • the pre-processing execution unit 56 executes various pre-processing (step S6: pre-processing execution step).
  • step S6 pre-processing execution step
  • step S7 alteration state identification step
  • the display control unit 58 causes the display means 2 to display the analysis results such as the discrimination result of the living cell and the identification result of the altered state (step S8: analysis result display step). As a result, the user can easily grasp the analysis result.
  • step S3 and steps S5 to S8 described above are not essential, and are appropriately executed as necessary or according to the user's choice.
  • the living cell analysis device 1A the living cell analysis system 100A, the living cell analysis program 1a, and the first embodiment of the living cell analysis method according to the present invention as described above, the following effects are obtained.
  • Living cells contained in the entire pathological specimen 10 can be analyzed automatically and with high accuracy.
  • the altered state of living cells can be identified automatically and with high accuracy.
  • Various alteration states in living cells can be identified.
  • the stand-alone type in which the biological cell analyzer 1A functions independently has been described.
  • the feature of the second embodiment is that the living cell analysis server 7 has the above-mentioned function of discriminating living cells and the function of discriminating the altered state.
  • the living cell analysis system 100B of the second embodiment mainly analyzes the living cells in addition to the stage 11, the hyperspectral camera 12, and the living cell analysis device 1B. It has a server 7.
  • the biological cell analysis device 1B has a display means 2, an input means 3, a storage means 4, an arithmetic processing means 5, and a communication means 6 separately.
  • the arithmetic processing means 5 of the biological cell analysis apparatus 1B functions as the whole scan instruction unit 51, the hyperspectral data acquisition unit 52, the input reception unit 53, and the display control unit 58, as in the first embodiment. Separately, it functions as a device-side communication unit 59.
  • the display means 2, the input means 3, and the storage means 4 are not shown.
  • the living cell analysis server 7 has a storage means 4, an arithmetic processing means 5, and a communication means 6.
  • the storage means 4 of the living cell analysis server 7 is the same as the storage means 4 possessed by the living cell analysis device 1A of the first embodiment.
  • the arithmetic processing means 5 of the biological cell analysis server 7 executes the biological cell analysis program 1b to execute the biological cell discrimination unit 54, the spectral information extraction unit 55, the pretreatment execution unit 56, and the alteration state identification unit. In addition to functioning as 57, it separately functions as a server-side communication unit 60.
  • the communication means 6 is configured by a wireless communication interface corresponding to a carrier network standard such as a 5th generation mobile communication network (5G) and a wireless LAN standard such as Wi-Fi (registered trademark).
  • a carrier network standard such as a 5th generation mobile communication network (5G) and a wireless LAN standard such as Wi-Fi (registered trademark).
  • the living cell analysis device 1B and the living cell analysis server 7 are connected to each other so as to be capable of data communication.
  • the standard of the communication means 6 is not limited to wireless communication, and may be wired communication.
  • the device-side communication unit 59 transmits hyperspectral data and information on the learned model selected by the user to the biological cell analysis server 7 via the communication means 6, and also performs biological cell analysis. Receive the analysis result from the server 7.
  • the server-side communication unit 60 receives hyperspectral data and learned model information from the biological cell analysis device 1B via the communication means 6, and analyzes the biological cell analysis device 1B. Send the result.
  • step S1 Hyperspectral data transmission step
  • the server-side communication unit 60 receives hyperspectral data and learned model information from the biological cell analysis device 1B via the communication means 6 (step S12: hyperspectral data reception step). ). After that, in the living cell analysis server 7, the processes from step S4 to step S7 are executed, and when an analysis result consisting of the discrimination result of the living cell and the identification result of the altered state is obtained, the server side communication unit 60 performs the analysis result. Is transmitted to the living cell analysis device 1B (step S13: analysis result transmission step).
  • the device-side communication unit 59 receives the analysis result (step S14: analysis result receiving step), and displays the analysis result on the display means 2 (step S8).
  • the living cell analysis device 1B can be used on the WEB browser, and anyone, anywhere, can easily obtain the analysis result of the living cell. Can be done.
  • the biological cell analysis server 7 executes the analysis process having a large processing load, the analysis speed can be improved and a large amount of hyperspectral data can be analyzed.
  • Example 1 an experiment was conducted to confirm the analysis performance of living cells using the living cell analysis device 1A of the first embodiment described above.
  • a pathological specimen 10 for cytopathology obtained by scraping the lungs as a living cell was prepared, and a pathologist marked the cell mass of cancer cells in advance.
  • the RGB image of the pathological specimen 10 is shown in FIG. 6 (a).
  • the circle mark in the center of the RGB image is attached by the pathologist.
  • the wavelengths of red (R): 700 [nm], green (G): 545 [nm], and blue (B): 440 [nm] of the hyperspectral data are selected.
  • the living cells contained in the entire pathological specimen 10 can be analyzed automatically and with high accuracy, and whether or not the living cells are altered is automatically and highly determined. It was shown to be accurately identified.
  • the living cell analysis device, the living cell analysis system, the living cell analysis program, and the living cell analysis method according to the present invention are not limited to the above-described embodiments, and can be appropriately changed.
  • the pathological specimen 10 for cytopathology which has been difficult to diagnose in the past, is used.
  • the analysis target is not limited to the pathological specimen 10 for cytopathology, and the pathological specimen 10 for histological diagnosis can also be analyzed.
  • a learned cell discrimination model is prepared for each body part in order to improve the cell discrimination accuracy. Further, in order to improve the identification accuracy of the altered state, a trained altered state identification model is prepared for each of the various altered states.
  • the configuration is not limited to this, and when only living cells in a specific site are discriminated and only a specific altered state is identified, a learned model corresponding to the site or the altered state is prepared. It is sufficient, and it is not necessary to separately accept the selection by the input receiving unit 53.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Organic Chemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Biochemistry (AREA)
  • Zoology (AREA)
  • Wood Science & Technology (AREA)
  • Analytical Chemistry (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Pathology (AREA)
  • Molecular Biology (AREA)
  • Genetics & Genomics (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Theoretical Computer Science (AREA)
  • Medicinal Chemistry (AREA)
  • Biotechnology (AREA)
  • Microbiology (AREA)
  • Biophysics (AREA)
  • Food Science & Technology (AREA)
  • Urology & Nephrology (AREA)
  • General Engineering & Computer Science (AREA)
  • Hematology (AREA)
  • Sustainable Development (AREA)
  • Mathematical Physics (AREA)
  • Virology (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Toxicology (AREA)
  • Cell Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
  • Image Analysis (AREA)

Abstract

Le problème décrit par la présente invention est de fournir un dispositif, un système, un programme et un procédé d'analyse de cellules vivantes qui peuvent analyser automatiquement et avec une grande précision une cellule vivante incluse dans un échantillon pathologique entier. La solution selon l'invention porte sur un dispositif d'analyse de cellules vivantes (1A) qui analyse une cellule vivante en utilisant des données hyperspectrales, le dispositif comprenant : une unité d'instruction de balayage complet (51) qui amène un étage (11), sur lequel un échantillon pathologique comprenant une cellule vivante est monté, et une caméra hyperspectrale (12) à être relativement déplacés et l'ensemble d'échantillon pathologique à être balayé ; une unité d'acquisition de données hyperspectrales (52) qui acquiert des données hyperspectrales de l'échantillon pathologique entier à partir de la caméra hyperspectrale (12) ; et une unité de détermination de cellule vivante qui détermine, sur la base des données hyperspectrales, un pixel correspondant à la cellule vivante à partir d'une image bidimensionnelle de l'échantillon pathologique entier au moyen d'un apprentissage automatique.
PCT/JP2021/022892 2020-06-17 2021-06-16 Dispositif, système, programme et procédé d'analyse de cellules vivantes Ceased WO2021256514A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2022531881A JP7427289B2 (ja) 2020-06-17 2021-06-16 生体細胞解析装置、生体細胞解析システム、生体細胞解析プログラムおよび生体細胞解析方法

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2020-104850 2020-06-17
JP2020104850 2020-06-17

Publications (1)

Publication Number Publication Date
WO2021256514A1 true WO2021256514A1 (fr) 2021-12-23

Family

ID=79268019

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/022892 Ceased WO2021256514A1 (fr) 2020-06-17 2021-06-16 Dispositif, système, programme et procédé d'analyse de cellules vivantes

Country Status (2)

Country Link
JP (1) JP7427289B2 (fr)
WO (1) WO2021256514A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023042647A1 (fr) * 2021-09-17 2023-03-23 シンクサイト株式会社 Procédé de génération de modèle de classification, procédé de classification de particules, programme informatique et dispositif de traitement d'informations
WO2023042646A1 (fr) * 2021-09-17 2023-03-23 シンクサイト株式会社 Procédé de génération de modèle de classification, procédé de détermination de particule, programme informatique et dispositif de traitement d'informations
WO2023248956A1 (fr) * 2022-06-20 2023-12-28 国立大学法人東京工業大学 Dispositif d'imagerie à bande photonique, procédé d'imagerie à bande photonique et programme

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017203637A (ja) * 2016-05-09 2017-11-16 住友電気工業株式会社 腫瘍細胞検出方法及び腫瘍細胞検出装置
JP2018205373A (ja) * 2017-05-31 2018-12-27 オリンパス株式会社 顕微鏡
US20190226991A1 (en) * 2018-01-22 2019-07-25 Verily Life Sciences Llc High-throughput hyperspectral imaging systems
WO2019178561A2 (fr) * 2018-03-16 2019-09-19 The United States Of America, As Represented By The Secretary, Department Of Health & Human Services Utilisation de réseaux neuronaux et/ou d'apprentissage machine pour valider des cellules souches et leurs dérivés pour une utilisation dans la thérapie cellulaire, la découverte de médicaments et le diagnostic
WO2019181845A1 (fr) * 2018-03-19 2019-09-26 一般財団法人未来科学研究所 Dispositif d'analyse de tissu biologique, programme d'analyse de tissu biologique et méthode d'analyse de tissu biologique

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017203637A (ja) * 2016-05-09 2017-11-16 住友電気工業株式会社 腫瘍細胞検出方法及び腫瘍細胞検出装置
JP2018205373A (ja) * 2017-05-31 2018-12-27 オリンパス株式会社 顕微鏡
US20190226991A1 (en) * 2018-01-22 2019-07-25 Verily Life Sciences Llc High-throughput hyperspectral imaging systems
WO2019178561A2 (fr) * 2018-03-16 2019-09-19 The United States Of America, As Represented By The Secretary, Department Of Health & Human Services Utilisation de réseaux neuronaux et/ou d'apprentissage machine pour valider des cellules souches et leurs dérivés pour une utilisation dans la thérapie cellulaire, la découverte de médicaments et le diagnostic
WO2019181845A1 (fr) * 2018-03-19 2019-09-26 一般財団法人未来科学研究所 Dispositif d'analyse de tissu biologique, programme d'analyse de tissu biologique et méthode d'analyse de tissu biologique

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HIZUKURI, AKIYOSHI ET AL.: "Cell detection and state classification using two-step machine learning", PROCEEDINGS OF VIEW VISION TECHNOLOGY PRACTICAL USE WORKSHOP, 8 December 2016 (2016-12-08) *
PEñARANDA FRANCISCO; NARANJO VALERY; LLOYD GAVIN R.; KASTL LENA; KEMPER BJöRN; SCHNEKENBURGER JüRGEN; NALLALA JAYAK: "Discrimination of skin cancer cells using Fourier transform infrared spectroscopy", COMPUTERS IN BIOLOGY AND MEDICINE, NEW YORK, NY, US, vol. 100, 28 June 2018 (2018-06-28), US , pages 50 - 61, XP085439079, ISSN: 0010-4825, DOI: 10.1016/j.compbiomed.2018.06.023 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023042647A1 (fr) * 2021-09-17 2023-03-23 シンクサイト株式会社 Procédé de génération de modèle de classification, procédé de classification de particules, programme informatique et dispositif de traitement d'informations
WO2023042646A1 (fr) * 2021-09-17 2023-03-23 シンクサイト株式会社 Procédé de génération de modèle de classification, procédé de détermination de particule, programme informatique et dispositif de traitement d'informations
WO2023248956A1 (fr) * 2022-06-20 2023-12-28 国立大学法人東京工業大学 Dispositif d'imagerie à bande photonique, procédé d'imagerie à bande photonique et programme

Also Published As

Publication number Publication date
JPWO2021256514A1 (fr) 2021-12-23
JP7427289B2 (ja) 2024-02-05

Similar Documents

Publication Publication Date Title
Han et al. Timely detection of skin cancer: an AI-based approach on the basis of the integration of echo state network and adapted seasons optimization algorithm
Sun et al. Diagnosis of cholangiocarcinoma from microscopic hyperspectral pathological dataset by deep convolution neural networks
CN111448569B (zh) 存储和检索数字病理学分析结果的方法
JP6366556B2 (ja) 生物学的試片をスペクトル画像により分析する方法
Grigoroiu et al. Deep learning applied to hyperspectral endoscopy for online spectral classification
JP7427289B2 (ja) 生体細胞解析装置、生体細胞解析システム、生体細胞解析プログラムおよび生体細胞解析方法
Xie et al. Interpretable classification from skin cancer histology slides using deep learning: A retrospective multicenter study
GB2466442A (en) A system to analyze a sample on a slide using Raman spectroscopy on an identified area of interest
KR20140043128A (ko) 스펙트럼 이미징에 의해 생물학적 표본을 분석하는 방법
AU2015265811A1 (en) An image processing method and system for analyzing a multi-channel image obtained from a biological tissue sample being stained by multiple stains
JP7693112B2 (ja) 画像処理方法、装置、機器、及びプログラム
JP6392476B1 (ja) 生体組織解析装置および生体組織解析プログラム
KR20160012114A (ko) 스펙트럼 이미지화에 의해 생물학적 시료를 분석하기 위한 방법 및 시스템
Ortega et al. Hyperspectral database of pathological in-vitro human brain samples to detect carcinogenic tissues
JP2025114735A (ja) 多重免疫蛍光イメージングを使用する組織学的染色のデジタル合成
Liu et al. Using hyperspectral imaging automatic classification of gastric cancer grading with a shallow residual network
US20240393248A1 (en) Information processing device, biological sample observation system, and image generation method
Courtenay et al. Using computational learning for non-melanoma skin cancer and actinic keratosis near-infrared hyperspectral signature classification
Youssef et al. A novel visible and near-infrared hyperspectral imaging platform for automated breast-cancer detection
KR20230063147A (ko) 다단계 특징 분석을 사용한 전립선 조직의 효율적인 경량 cnn과 앙상블 머신 러닝 분류 방법 및 시스템
JP2024061696A (ja) 医療用画像解析装置、医療用画像解析方法及び医療用画像解析システム
CN115176139A (zh) 用于调整训练门以适应流式细胞仪数据的方法和系统
Liu et al. Signal to noise ratio quantifies the contribution of spectral channels to classification of human head and neck tissues ex vivo using deep learning and multispectral imaging
de Lucena et al. Detection of tumoral epithelial lesions using hyperspectral imaging and deep learning
Grechkin et al. VGG convolutional neural network classification of hyperspectral images of skin neoplasms

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: 21825675

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022531881

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21825675

Country of ref document: EP

Kind code of ref document: A1